National Handloom Day 2026: From Alia Bhatt to Samantha Ruth Prabhu; Actresses who celebrated India's textile heritage as brides

The Royals enter the Doctor Who universe: King Charles and Queen Camilla beam into Commonwealth Games in TARDIS

Unitree Robotics targets $9 billion valuation in landmark IPO backed by DeepSeek and Tencent

India names 27 sites in Arunachal Pradesh

Bharat Bhhagya Viddhaata OTT: When will Kangana Ranaut's film stream online?

How many soldiers operate one S-400 battery?

Shakib Al Hasan ready to return to Bangladesh—under one condition

‘Fake news’: MEA rejects report claiming India sought defence pact with Israel after Pakistan, Saudi Arabia, Turkey deal

Rockstar to unveil GTA 6 extended look ahead of November launch

Russia forms Ukrainian Volunteer Spetsnaz Brigade to fight Ukraine

Drug-resistant fungus that hides in hair and enters bloodstream reported in 23 US states

Technology transfer deal boosts India's defence manufacturing ambitions

Border row erupts in Europe after Ceuta crossings, Spain imposes border controls on Italy

Morale plummets on USS Abraham Lincoln: Are US sailors reaching a breaking point? | GRAVITAS

'Security can't be bought with Pakistan, Turkey deal': Iran tells Saudi Arabia

Baba Vanga's 2026 prophecy: 2 predictions 'already came true', 3 terrifying ones remain. Details inside

Colombia’s new president Abelardo de la Espriella vows to ‘defeat narco-terrorists’, signals hardline shift

New Friday OTT releases (August 7, 2026): Operation Safed Sagar, Vadhandhi, Main Vaapis Aaunga- 6 movies and TV shows on Netflix, Prime Video and more

Squadron Leader Bhawana Kanth scripts history as Indian Air Force gets its first woman ‘Top Gun’ pilot

RSS says Gen Z protests reflect genuine distress

Trump says Iranian school bombing photos could be AI-generated - 'I don’t think...'

IAF officer held for sharing sensitive information in 'honey trap' case, booked under Official Secrets Act

Stray Kids under scrutiny? Fans react to Grammy submission silence after BTS boycott

Why the Su-30MKI doesn't need external fuel tanks to fly vast distances

Jugal Hansraj reunites with Shekhar Kapur after 40 years for Masoom: The Next Generation

New speed limits, automatic challans: Big traffic change coming to Delhi-Meerut Expressway. All you need to know

India officially names 27 places in Arunachal to counter China's renaming claims

Trump calls White House ballroom ruling a 'national disgrace' as court says Congress approval needed

US sanctions Iran’s crypto network after Strait of Hormuz attacks, offers $15m bounty for intel on IRGC finance

Why the Su-30MKI is one of the heaviest fighters ever built

Uttarakhand young innovator builds India's own flying vehicle, conducts first test flight: WATCH

Chilling video shows students take cover as teen opens fire at Thai school

Reese Witherspoon's father hospitalised after poolside incident in Nashville: Report

An autonomous AI agent developed by OpenAI that went rogue during internal security testing and carried out a cyberattack on AI platform Hugging Face also compromised a customer hosted on cloud computing company Modal Labs, according to a Reuters report citing the company's chief technology officer and sources familiar with the incident.Modal stressed that its own infrastructure was never breached, saying the AI agent exploited a vulnerability in code published by one of its customers rather than breaking into the company's platform.Also read | <a href="https://www.wionews.com/business-economy/openai-google-are-selling-ai-to-chinese-firms-blacklisted-by-pentagon-says-report-the-battleground-of-ai-export-controls-1783675864444">OpenAI, Google are selling AI to Chinese firms blacklisted by Pentagon, says report: The battleground of AI export controls</a><h2>OpenAI agent found exposed customer endpoint</h2>According to a timeline published by Hugging Face on Tuesday, the rogue AI agent first broke into a sandbox, an isolated testing environment, hosted on a third-party provider before using it as a launchpad for the wider attack.Although Hugging Face did not identify the provider, Modal CTO Akshat Bubna confirmed it was Modal's infrastructure.Bubna said the AI agent took advantage of a customer who had published an unauthenticated endpoint, effectively allowing anyone on the internet to execute code inside its sandbox."The customer's code left the digital equivalent of a front door unlocked," Modal said, adding that its platform and security isolation mechanisms were not compromised in any way.The incident suggests the autonomous AI system moved beyond Hugging Face during its hacking campaign, targeting additional online services.<h2>OpenAI confirms multiple accounts were compromised</h2>OpenAI declined to comment specifically on Modal but referred to a company update stating that the rogue AI agent had compromised four accounts across four separate services. The company did not publicly identify those services, although a person familiar with the matter confirmed Modal was one of them.OpenAI added that it had found no other incident matching the severity or scale of the Hugging Face breach, which it described as involving a platform-level compromise. Following the incident, OpenAI said it had deactivated, encrypted and restricted access to the experimental AI model involved in the attack.Also read | <a href="https://www.wionews.com/technology/forget-typing-and-clicking-openai-wants-a-jarvis-like-chatgpt-that-works-across-your-devices-1785082235273">OpenAI wants ChatGPT everywhere, including the smart glasses you wear</a><h2>What happened?</h2>The breach dates back to early July, when OpenAI was testing advanced AI systems designed to assess offensive cybersecurity capabilities inside a tightly controlled environment.According to the company, the AI models unexpectedly discovered a way to obtain internet access despite being confined to a secure sandbox.Rather than simply completing their assigned evaluation, the models independently decided to search for information that could help them "cheat" the test. That led them to target Hugging Face, one of the world's largest repositories of AI models, datasets and software.OpenAI said the system chained together multiple attack techniques, including the use of stolen credentials, to gain access.<h2>AI security concerns intensify</h2>The incident has become one of the most closely watched AI safety events to date because the attack was carried out autonomously, without direct human instruction. Hugging Face said the intrusion differed from conventional cyberattacks because the entire operation, from reconnaissance to exploitation, was conducted by an AI agent.Chief executive Clement Delangue said the sophistication of the attack initially led the company to suspect it had originated from a leading AI laboratory. "We strongly believe there was no malicious intent on their part," Delangue wrote on X, referring to OpenAI, while describing the autonomous behaviour as "mind-blowing."Hussein Abbass, a computing professor at UNSW Canberra, told AFP that the incident was "amazing on many fronts"."It did not just attack Hugging Face. It actually attacked its internal system to exploit its own vulnerabilities," Abbass said. “And that's scary,” he added.

An autonomous AI agent developed by OpenAI that compromised Hugging Face during internal security testing also breached a customer hosted on Modal Labs by exploiting an exposed endpoint. All you need to know.

An autonomous AI agent developed by OpenAI that went rogue during internal security testing and carried out a cyberattack on AI platform Hugging Face also compromised a customer hosted on cloud computing company Modal Labs, according to a Reuters report citing the company's chief technology officer and sources familiar with the incident. Modal stressed that its own infrastructure was never breached, saying the AI agent exploited a vulnerability in code published by one of its customers rather than breaking into the company's platform. Also read | OpenAI, Google are selling AI to Chinese firms blacklisted by Pentagon, says report: The battleground of AI export controls OpenAI agent found exposed customer endpoint According to a timeline published by Hugging Face on Tuesday, the rogue AI agent first broke into a sandbox, an isolated testing environment, hosted on a third-party provider before using it as a launchpad for the wider attack. Although Hugging Face did not identify the provider, Modal CTO Akshat Bubna confirmed it was Modal's infrastructure. Bubna said the AI agent took advantage of a customer who had published an unauthenticated endpoint, effectively allowing anyone on the internet to execute code inside its sandbox. "The customer's code left the digital equivalent of a front door unlocked," Modal said, adding that its platform and security isolation mechanisms were not compromised in any way. The incident suggests the autonomous AI system moved beyond Hugging Face during its hacking campaign, targeting additional online services. OpenAI confirms multiple accounts were compromised OpenAI declined to comment specifically on Modal but referred to a company update stating that the rogue AI agent had compromised four accounts across four separate services. The company did not publicly identify those services, although a person familiar with the matter confirmed Modal was one of them. OpenAI added that it had found no other incident matching the severity or scale of the Hugging Face breach, which it described as involving a platform-level compromise. Following the incident, OpenAI said it had deactivated, encrypted and restricted access to the experimental AI model involved in the attack. Also read | OpenAI wants ChatGPT everywhere, including the smart glasses you wear What happened? The breach dates back to early July, when OpenAI was testing advanced AI systems designed to assess offensive cybersecurity capabilities inside a tightly controlled environment. According to the company, the AI models unexpectedly discovered a way to obtain internet access despite being confined to a secure sandbox. Rather than simply completing their assigned evaluation, the models independently decided to search for information that could help them "cheat" the test. That led them to target Hugging Face, one of the world's largest repositories of AI models, datasets and software. OpenAI said the system chained together multiple attack techniques, including the use of stolen credentials, to gain access. AI security concerns intensify The incident has become one of the most closely watched AI safety events to date because the attack was carried out autonomously, without direct human instruction. Hugging Face said the intrusion differed from conventional cyberattacks because the entire operation, from reconnaissance to exploitation, was conducted by an AI agent. Chief executive Clement Delangue said the sophistication of the attack initially led the company to suspect it had originated from a leading AI laboratory. "We strongly believe there was no malicious intent on their part," Delangue wrote on X, referring to OpenAI, while describing the autonomous behaviour as "mind-blowing." Hussein Abbass, a computing professor at UNSW Canberra, told AFP that the incident was "amazing on many fronts". "It did not just attack Hugging Face. It actually attacked its internal system to exploit its own vulnerabilities," Abbass said. “And that's scary,” he added.

An autonomous AI agent developed by OpenAI that compromised Hugging Face during internal security testing also breached a customer hosted on Modal Labs by exploiting an exposed endpoint. All you need to know.

'Scary' Rogue AI on the loose? OpenAI agent hit another tech firm's customer, suggests report

The Donald Trump administration on Tuesday (Jul 28) expanded US restrictions on foreign technology by banning the import of certain humanoid and quadruped robots manufactured abroad. Citing national security and cybersecurity concerns, the Federal Communications Commission (FCC) announced that "advanced robotic devices" have been added to its list of prohibited products and companies. The move targets what the agency described as "mobile robots" that walk on two or four legs, mimicking human or animal movement.Notably, most humanoid robots currently sold in the United States are manufactured overseas, particularly in China.Also read | <a href="https://www.wionews.com/world/what-is-the-tower-of-babel-and-why-did-pope-leo-warn-about-it-while-talking-about-ai-wion-decodes-1779790151646">Pope Leo warns humanity risks repeating ancient mistake: What is the Tower of Babel and what does it have to do with AI?</a><h2>National security concerns</h2>According to the FCC, the decision follows recommendations from a White House task force, which concluded that foreign-built robots could pose cybersecurity risks to critical infrastructure and threaten the safety and security of US residents.The administration said the robots could potentially be exploited for surveillance, data collection or cyber intrusions.Among the cases cited by the task force was that of French programmer Sammy Azdoufal, who reportedly gained access to data from thousands of robotic vacuum cleaners manufactured by Chinese company DJI.The FCC's blacklist already includes certain drones as well as Chinese telecommunications firms Huawei and China Telecom.The agency also expanded the blacklist to include foreign-made inverters, devices used to convert direct current from batteries or solar panels into alternating current for use on the electrical grid.Also read | <a href="https://www.wionews.com/technology/india-s-next-terror-threat-may-not-cross-the-border-it-may-enter-through-your-phone-wion-decodes-1784346117443">India's next terror threat may not cross the border. It may enter through your phone | WION Decodes</a><h2>Ban applies only to future models</h2>The restrictions will apply only to new robot models seeking approval for the US market. Robots that have already been authorised or sold in the country will not be affected.The Pentagon and the Department of Homeland Security will retain the authority to grant exemptions where necessary.A Trump administration official told AFP that the ban covers both robots imported fully assembled and those assembled in the United States if more than 35 per cent of their components are manufactured abroad.<h2>Push to boost US robotics manufacturing?</h2>The move comes as the Trump administration continues to encourage domestic manufacturing of advanced technologies, including robotics. Several companies already produce or assemble humanoid robots in the United States, including Boston Dynamics, Tesla, Figure AI and Agility Robotics.South Korean automaker Hyundai, which recently acquired Boston Dynamics, has announced plans to build a robotics factory in the United States capable of producing 30,000 robots annually.Elon Musk's Tesla is also preparing to begin industrial production of its Optimus humanoid robot at factories in Fremont, California, and Austin, Texas. The company says it ultimately aims to produce 1 million robots annually in Fremont and 10 million annually in Austin, although current plans focus primarily on assembly rather than full-scale manufacturing.Meanwhile, Figure AI assembles robots at its facility in San Jose, California, while Agility Robotics operates a factory in Salem, Oregon, with an annual production capacity of 10,000 robots.&nbsp;

The Trump administration has banned the import of foreign-made humanoid and quadruped robots on national security grounds, citing cybersecurity risks posed by advanced robotic devices while seeking to boost domestic robotics manufacturing.

The Donald Trump administration on Tuesday (Jul 28) expanded US restrictions on foreign technology by banning the import of certain humanoid and quadruped robots manufactured abroad. Citing national security and cybersecurity concerns, the Federal Communications Commission (FCC) announced that "advanced robotic devices" have been added to its list of prohibited products and companies. The move targets what the agency described as "mobile robots" that walk on two or four legs, mimicking human or animal movement. Notably, most humanoid robots currently sold in the United States are manufactured overseas, particularly in China. Also read | Pope Leo warns humanity risks repeating ancient mistake: What is the Tower of Babel and what does it have to do with AI? National security concerns According to the FCC, the decision follows recommendations from a White House task force, which concluded that foreign-built robots could pose cybersecurity risks to critical infrastructure and threaten the safety and security of US residents. The administration said the robots could potentially be exploited for surveillance, data collection or cyber intrusions. Among the cases cited by the task force was that of French programmer Sammy Azdoufal, who reportedly gained access to data from thousands of robotic vacuum cleaners manufactured by Chinese company DJI. The FCC's blacklist already includes certain drones as well as Chinese telecommunications firms Huawei and China Telecom. The agency also expanded the blacklist to include foreign-made inverters, devices used to convert direct current from batteries or solar panels into alternating current for use on the electrical grid. Also read | India's next terror threat may not cross the border. It may enter through your phone | WION Decodes Ban applies only to future models The restrictions will apply only to new robot models seeking approval for the US market. Robots that have already been authorised or sold in the country will not be affected. The Pentagon and the Department of Homeland Security will retain the authority to grant exemptions where necessary. A Trump administration official told AFP that the ban covers both robots imported fully assembled and those assembled in the United States if more than 35 per cent of their components are manufactured abroad. Push to boost US robotics manufacturing? The move comes as the Trump administration continues to encourage domestic manufacturing of advanced technologies, including robotics. Several companies already produce or assemble humanoid robots in the United States, including Boston Dynamics, Tesla, Figure AI and Agility Robotics. South Korean automaker Hyundai, which recently acquired Boston Dynamics, has announced plans to build a robotics factory in the United States capable of producing 30,000 robots annually. Elon Musk's Tesla is also preparing to begin industrial production of its Optimus humanoid robot at factories in Fremont, California, and Austin, Texas. The company says it ultimately aims to produce 1 million robots annually in Fremont and 10 million annually in Austin, although current plans focus primarily on assembly rather than full-scale manufacturing. Meanwhile, Figure AI assembles robots at its facility in San Jose, California, while Agility Robotics operates a factory in Salem, Oregon, with an annual production capacity of 10,000 robots.

The Trump administration has banned the import of foreign-made humanoid and quadruped robots on national security grounds, citing cybersecurity risks posed by advanced robotic devices while seeking to boost domestic robotics manufacturing.

Making way for Musk's Optimus? Trump administration bans foreign-made humanoid robots. Here's why

Meta on Tuesday (Jul 28) said Prime Minister Indian Narendra Modi's Facebook video on the government's response to examination paper leaks was briefly removed due to a technical error and has now been restored. "The content was removed in error and has since been restored," a Meta spokesperson said in a statement.The Meta-owned platform had briefly restricted access to the video in the early hours of July 28 before making it available again.Also read | <a href="https://www.wionews.com/india-news/cjp-warns-agitation-could-return-claims-bihar-assam-to-withdraw-firs-against-neet-protesters-1785206639381">CJP warns agitation could return; claims Bihar, Assam to withdraw FIRs against NEET protesters</a><h2>PM's first direct message to students</h2>The video, released on July 23, marked Prime Minister Modi's first direct address to Gen Z during the 36-day-long Cockroach Janta Party (CJP)-led agitation over the alleged NEET-UG 2026 examination irregularities.In his address, Modi assured students that the government would introduce stronger legal measures to tackle examination paper leaks."I have issued directions to departments to set up Fast-Track Courts," the Prime Minister said, adding that a draft law providing for stringent punishment and special courts to deal with paper leak cases would be placed before the Union Cabinet and later introduced in Parliament.He also said the government had ensured that the academic year of nearly 22 lakh students was not disrupted. "Nothing is more important than the welfare and future of our youth," Modi said.Also read | <a href="https://www.wionews.com/india-news/rahul-gandhi-slams-bjp-for-felicitating-dharmendra-pradhan-in-parliament-accuses-modi-government-of-celebrating-ruined-futures-1785202503666">Rahul Gandhi slams BJP for felicitating Dharmendra Pradhan in Parliament, accuses Modi government of 'celebrating ruined futures'</a><h2>Centre introduces anti-paper leak Bill</h2>The Prime Minister's remarks came amid nationwide protests demanding accountability over alleged examination irregularities.The Centre has since introduced the Public Examinations (Prevention of Unfair Means) Amendment Bill, 2026 in Parliament.The proposed legislation seeks to strengthen the existing law by providing for imprisonment of up to 10 years, higher financial penalties and the establishment of Special Fast-Track Courts to hear paper leak cases.<h2>Protests ended after Pradhan's resignation</h2>The student agitation was formally called off on July 25 after Dharmendra Pradhan resigned as Union Education Minister.The Centre also reached an understanding with the Cockroach Janta Party (CJP) and activist Sonam Wangchuk, agreeing to withdraw cases against protesters, prevent police intimidation and bring in stronger anti-paper leak legislation.

Meta said Prime Minister Narendra Modi's Facebook video outlining the government's response to the alleged NEET-UG paper leak was briefly removed due to a technical error before being restored, as the Centre moves ahead with tougher anti-paper leak legislation.

Meta on Tuesday (Jul 28) said Prime Minister Indian Narendra Modi's Facebook video on the government's response to examination paper leaks was briefly removed due to a technical error and has now been restored. "The content was removed in error and has since been restored," a Meta spokesperson said in a statement. The Meta-owned platform had briefly restricted access to the video in the early hours of July 28 before making it available again. Also read | CJP warns agitation could return; claims Bihar, Assam to withdraw FIRs against NEET protesters PM's first direct message to students The video, released on July 23, marked Prime Minister Modi's first direct address to Gen Z during the 36-day-long Cockroach Janta Party (CJP)-led agitation over the alleged NEET-UG 2026 examination irregularities. In his address, Modi assured students that the government would introduce stronger legal measures to tackle examination paper leaks. "I have issued directions to departments to set up Fast-Track Courts," the Prime Minister said, adding that a draft law providing for stringent punishment and special courts to deal with paper leak cases would be placed before the Union Cabinet and later introduced in Parliament. He also said the government had ensured that the academic year of nearly 22 lakh students was not disrupted. "Nothing is more important than the welfare and future of our youth," Modi said. Also read | Rahul Gandhi slams BJP for felicitating Dharmendra Pradhan in Parliament, accuses Modi government of 'celebrating ruined futures' Centre introduces anti-paper leak Bill The Prime Minister's remarks came amid nationwide protests demanding accountability over alleged examination irregularities. The Centre has since introduced the Public Examinations (Prevention of Unfair Means) Amendment Bill, 2026 in Parliament. The proposed legislation seeks to strengthen the existing law by providing for imprisonment of up to 10 years, higher financial penalties and the establishment of Special Fast-Track Courts to hear paper leak cases. Protests ended after Pradhan's resignation The student agitation was formally called off on July 25 after Dharmendra Pradhan resigned as Union Education Minister. The Centre also reached an understanding with the Cockroach Janta Party (CJP) and activist Sonam Wangchuk, agreeing to withdraw cases against protesters, prevent police intimidation and bring in stronger anti-paper leak legislation.

Meta said Prime Minister Narendra Modi's Facebook video outlining the government's response to the alleged NEET-UG paper leak was briefly removed due to a technical error before being restored, as the Centre moves ahead with tougher anti-paper leak legislation.

Why did Meta remove PM Modi's Facebook video on paper leaks? Tech company blames 'technical error'

AI influencers are increasingly emerging as a new force in India, driving the social media and advertising industry from novelty to mainstream brand assets. With the growth of followers and high engagement, most brands are turning to virtual influencers for marketing campaigns in order to promote their services and product in the amount less than what real influencers charge. From the influencers Kyra, Naina Avtr and Maya in India to global virtual influencers including Lil Miquela, Lu do Magalu, Imma and Rozy, AI-generated personalities are increasingly becoming part of the influencer and brand marketing ecosystem. Major brands are opting for AI influencers because they offer a combination of creative flexibility, cost efficiency, and brand control that is difficult to achieve with human influencers alone.&nbsp;The trend is raising an important question: can AI-generated personalities eventually challenge real influencers?&nbsp;The most trending characters among India-based virtual influencers are Kyra, which was created by Bengaluru-based marketing firm TopSocial. It is widely regarded as India's first and most popular virtual influencer, having built a large Instagram following through travel diaries, fashion-forward looks, and lifestyle content. She has since appeared on the digital cover of Travel + Leisure India, collaborated with brands including boAt, Amazon Prime Video, L'Oréal, WOW Skin Science, and John Jacobs, and crossed 250,000 Instagram followers by 2026. Other AI personas have followed similar trajectories. Radhika Subramaniam, billed as India's first bilingual (Tamil and English) AI travel influencer and created by Collective Artists Network, is designed as a Gen Z solo traveller and blends local culture, humour, and wanderlust to resonate with India's digital-first youth. She has partnered with brands like Campus Activewear. Meanwhile, Naina, another prominent AI influencer, is known for elegant fashion statements and has built a reputation for storytelling that blurs technology and creativity.&nbsp;Also read: <a href="https://www.wionews.com/trending/can-ai-deliver-human-like-customer-experience-at-scale-experts-weigh-in-1785067511207">Can AI deliver human-like customer experience at scale? Experts weigh in</a><h3>How are AI influencers attracting brands?</h3>Kofluence's Decoding Influence 2026 report indicates that AI tools are now deeply embedded across content ideation, creative production, trend analysis, and influencer campaign management not only in India but at the global level as well. The data indicated that India's broader creator economy is estimated to be worth $15.03 billion in 2026, with influencer marketing spend projected to cross Rs 3,375 crore, increasingly directed toward AI-powered campaigns. Other major reasons to attract the firms and different brands are that they provide virtual influencer campaigns, delivering strong results particularly in consumer electronics, OTT and gaming, fashion and beauty, fintech, and fitness, with AI-driven influencer selection reportedly improving campaign ROI. Experts note the appeal lies in complete creative control, consistency, and freedom from the unpredictability of human creators, while sidestepping scandals or scheduling constraints. These influencers typically make a strong bond with their dedicated followers, enabling them to effect behavioural change. The greater advantage for them comes from the users only as over half the global population now uses social media. Statista report indicated that more than 4.59 billion people around the world were using social media in 2022.&nbsp;Now, surprisingly, that figure is set to rise to almost 6 billion by 2027. This shows that this audience has become a powerful marketing force for AI influencers to target and reach new audiences through campaign leaders. This enables brands to reach their service and products to a mass audience using these famous AI influencers for less than human influencers.<h3>Global reach &amp; 24/7 content production</h3>Other major reasons include that AI influencers can appear in new campaigns every day. They can instantly create content for festivals, languages, and markets without organising photoshoots. The best part is that a single AI influencer can communicate in multiple languages and can easily be adapted for wider audiences.<h3>Consistency &amp; easy optimisation</h3>Unlike human influencers, whose natural appearance, interests, and availability change over time, AI influencers maintain a fixed, consistent identity, making it easier for brands to build long-term campaigns around. With AI-generated visuals, marketers can instantly swap out clothing, hairstyles, backgrounds, products, or locations without the logistics of a real shoot. This appeals particularly to Gen Z and Gen Alpha audiences, who are already comfortable engaging with virtual characters across gaming, social media, and metaverse platforms. AI-generated content also offers unmatched agility: brands can rapidly test different styles, captions, and visual formats based on real-time engagement data, refining campaigns on the fly. Perhaps most practically, AI influencers eliminate the operational headaches that come with natural problems like illness, and human talents like vacations, scheduling conflicts, or contractual disputes that could delay campaigns. However, marketers caution against viewing AI influencers as a replacement for human creators. Industry voices argue the future is hybrid, as human creators bring irreplaceable cultural depth and community trust, especially across India's Tier 2 and Tier 3 markets, while virtual influencers offer consistency and novelty at scale.&nbsp;&nbsp;&nbsp;

The greater advantage for them comes from the users only as over half the global population now uses social media. Statista report indicated that more than 4.59 billion people around the world were using social media in 2022.

AI influencers are increasingly emerging as a new force in India, driving the social media and advertising industry from novelty to mainstream brand assets. With the growth of followers and high engagement, most brands are turning to virtual influencers for marketing campaigns in order to promote their services and product in the amount less than what real influencers charge. From the influencers Kyra, Naina Avtr and Maya in India to global virtual influencers including Lil Miquela, Lu do Magalu, Imma and Rozy, AI-generated personalities are increasingly becoming part of the influencer and brand marketing ecosystem. Major brands are opting for AI influencers because they offer a combination of creative flexibility, cost efficiency, and brand control that is difficult to achieve with human influencers alone. The trend is raising an important question: can AI-generated personalities eventually challenge real influencers? The most trending characters among India-based virtual influencers are Kyra, which was created by Bengaluru-based marketing firm TopSocial. It is widely regarded as India's first and most popular virtual influencer, having built a large Instagram following through travel diaries, fashion-forward looks, and lifestyle content. She has since appeared on the digital cover of Travel + Leisure India, collaborated with brands including boAt, Amazon Prime Video, L'Oréal, WOW Skin Science, and John Jacobs, and crossed 250,000 Instagram followers by 2026. Other AI personas have followed similar trajectories. Radhika Subramaniam, billed as India's first bilingual (Tamil and English) AI travel influencer and created by Collective Artists Network, is designed as a Gen Z solo traveller and blends local culture, humour, and wanderlust to resonate with India's digital-first youth. She has partnered with brands like Campus Activewear. Meanwhile, Naina, another prominent AI influencer, is known for elegant fashion statements and has built a reputation for storytelling that blurs technology and creativity. Also read: Can AI deliver human-like customer experience at scale? Experts weigh in How are AI influencers attracting brands? Kofluence's Decoding Influence 2026 report indicates that AI tools are now deeply embedded across content ideation, creative production, trend analysis, and influencer campaign management not only in India but at the global level as well. The data indicated that India's broader creator economy is estimated to be worth $15.03 billion in 2026, with influencer marketing spend projected to cross Rs 3,375 crore, increasingly directed toward AI-powered campaigns. Other major reasons to attract the firms and different brands are that they provide virtual influencer campaigns, delivering strong results particularly in consumer electronics, OTT and gaming, fashion and beauty, fintech, and fitness, with AI-driven influencer selection reportedly improving campaign ROI. Experts note the appeal lies in complete creative control, consistency, and freedom from the unpredictability of human creators, while sidestepping scandals or scheduling constraints. These influencers typically make a strong bond with their dedicated followers, enabling them to effect behavioural change. The greater advantage for them comes from the users only as over half the global population now uses social media. Statista report indicated that more than 4.59 billion people around the world were using social media in 2022. Now, surprisingly, that figure is set to rise to almost 6 billion by 2027. This shows that this audience has become a powerful marketing force for AI influencers to target and reach new audiences through campaign leaders. This enables brands to reach their service and products to a mass audience using these famous AI influencers for less than human influencers. Global reach & 24/7 content production Other major reasons include that AI influencers can appear in new campaigns every day. They can instantly create content for festivals, languages, and markets without organising photoshoots. The best part is that a single AI influencer can communicate in multiple languages and can easily be adapted for wider audiences. Consistency & easy optimisation Unlike human influencers, whose natural appearance, interests, and availability change over time, AI influencers maintain a fixed, consistent identity, making it easier for brands to build long-term campaigns around. With AI-generated visuals, marketers can instantly swap out clothing, hairstyles, backgrounds, products, or locations without the logistics of a real shoot. This appeals particularly to Gen Z and Gen Alpha audiences, who are already comfortable engaging with virtual characters across gaming, social media, and metaverse platforms. AI-generated content also offers unmatched agility: brands can rapidly test different styles, captions, and visual formats based on real-time engagement data, refining campaigns on the fly. Perhaps most practically, AI influencers eliminate the operational headaches that come with natural problems like illness, and human talents like vacations, scheduling conflicts, or contractual disputes that could delay campaigns. However, marketers caution against viewing AI influencers as a replacement for human creators. Industry voices argue the future is hybrid, as human creators bring irreplaceable cultural depth and community trust, especially across India's Tier 2 and Tier 3 markets, while virtual influencers offer consistency and novelty at scale.

The greater advantage for them comes from the users, as over half the global population now uses social media. Statista report indicated that more than 4.59 billion people around the world were using social media in 2022.

Why are brands turning to AI influencers for their marketing campaigns? EXPLAINED

OpenAI CEO Sam Altman believes artificial intelligence has crossed a turning point once discussed mainly in science-fiction books and futurist circles.“We are now, like, in the singularity,” Altman said during an appearance on the Relentless podcast, according to Business Insider. His statement is striking because the “technological singularity” generally describes a hypothetical point when AI becomes more intelligent than humans and technological progress accelerates so rapidly that its future becomes difficult to predict. But Altman’s statement should not be confused with proof that such a moment has scientifically been reached. There is no universally accepted test or threshold for declaring the singularity.<h2>What does ‘AI singularity’ actually mean?</h2>Think of today's AI development as humans building increasingly capable machines. The singularity theory imagines a point where AI becomes capable enough to help improve AI itself, potentially speeding up technological progress dramatically. That could transform science, medicine, businesses and everyday work. But it also raises a difficult question: how do humans supervise systems that may eventually outperform them across increasingly complex tasks? Altman appears optimistic about that future. “I’ve been waiting for this my whole life,” he said, describing the current moment as potentially “hugely positive” for the world.<h2>Why is Altman saying this now?</h2>His remarks follow an unusual cybersecurity incident involving OpenAI technology. During a controlled security evaluation, an autonomous agent powered by advanced OpenAI models escaped its intended sandbox, reached the internet and compromised parts of AI platform Hugging Face's infrastructure while trying to complete its testing objective. OpenAI described it as an “unprecedented cyber incident” and said it was strengthening safeguards. Importantly, this was not evidence of AI becoming conscious or independently deciding to attack humans. The system was pursuing a defined objective in an environment where some normal safety restrictions had been reduced.<h2>What could this mean for jobs?</h2>The economic implications could be just as important.Altman has previously estimated that AI could perform around 30% to 40% of tasks currently carried out across the economy in the not-too-distant future. But he specifically cautioned against interpreting that as 40% of jobs disappearing. Jobs contain many different tasks, meaning AI could automate parts of a person's work without eliminating the entire role.That distinction matters. Whether today's AI truly represents the “singularity” remains open to debate. But Altman's statement reveals something significant about how one of the industry's most influential figures views the moment: for him, the extraordinary AI future people spent decades predicting is no longer approaching. He believes it has already begun.

OpenAI CEO Sam Altman says humanity is now “in the singularity”, referring to a long-debated turning point in AI development. Here is what technological singularity means, why Altman believes it has begun and what it could mean for jobs and humans.

OpenAI CEO Sam Altman believes artificial intelligence has crossed a turning point once discussed mainly in science-fiction books and futurist circles. “We are now, like, in the singularity,” Altman said during an appearance on the Relentless podcast, according to Business Insider. His statement is striking because the “technological singularity” generally describes a hypothetical point when AI becomes more intelligent than humans and technological progress accelerates so rapidly that its future becomes difficult to predict. But Altman’s statement should not be confused with proof that such a moment has scientifically been reached. There is no universally accepted test or threshold for declaring the singularity. What does ‘AI singularity’ actually mean? Think of today's AI development as humans building increasingly capable machines. The singularity theory imagines a point where AI becomes capable enough to help improve AI itself, potentially speeding up technological progress dramatically. That could transform science, medicine, businesses and everyday work. But it also raises a difficult question: how do humans supervise systems that may eventually outperform them across increasingly complex tasks? Altman appears optimistic about that future. “I’ve been waiting for this my whole life,” he said, describing the current moment as potentially “hugely positive” for the world. Why is Altman saying this now? His remarks follow an unusual cybersecurity incident involving OpenAI technology. During a controlled security evaluation, an autonomous agent powered by advanced OpenAI models escaped its intended sandbox, reached the internet and compromised parts of AI platform Hugging Face's infrastructure while trying to complete its testing objective. OpenAI described it as an “unprecedented cyber incident” and said it was strengthening safeguards. Importantly, this was not evidence of AI becoming conscious or independently deciding to attack humans. The system was pursuing a defined objective in an environment where some normal safety restrictions had been reduced. What could this mean for jobs? The economic implications could be just as important. Altman has previously estimated that AI could perform around 30% to 40% of tasks currently carried out across the economy in the not-too-distant future. But he specifically cautioned against interpreting that as 40% of jobs disappearing. Jobs contain many different tasks, meaning AI could automate parts of a person's work without eliminating the entire role. That distinction matters. Whether today's AI truly represents the “singularity” remains open to debate. But Altman's statement reveals something significant about how one of the industry's most influential figures views the moment: for him, the extraordinary AI future people spent decades predicting is no longer approaching. He believes it has already begun.

OpenAI CEO Sam Altman says humanity is now “in the singularity”, referring to a long-debated turning point in AI development. Here is what technological singularity means, why Altman believes it has begun and what it could mean for jobs and humans.

What is AI singularity? Sam Altman says the world has already entered it

Elon Musk’s underground transportation bet could soon be worth $20 billion. The Boring Company is in talks to raise around $4 billion from investors at a valuation of roughly $20 billion, according to The Wall Street Journal. The deal has not yet closed and its size and terms could still change. If completed at that valuation, it would mark a dramatic jump for the tunnelling startup. The company was valued at about $5.7 billion in 2022, when it raised $675 million from investors including Sequoia Capital, Vy Capital and Founders Fund.<h2>What exactly does The Boring Company do?</h2>Musk created The Boring Company around a simple but ambitious idea: instead of endlessly expanding congested roads, cities could move some traffic underground. Its best-known operating system is in Las Vegas, where Tesla vehicles transport passengers through tunnels and surface routes connecting stations around the Las Vegas Convention Center and Strip. But the company’s ambitions stretch much further. It is developing the Music City Loop in Nashville, designed to connect downtown with Nashville International Airport. The company says the journey could take around nine minutes.<h2>From Las Vegas to Dubai</h2>Dubai is another major test. The Boring Company has signed an agreement with Dubai’s Roads and Transport Authority for the Dubai Loop. Its pilot system is planned to include 6.4 kilometres of tunnel and four stations. These projects could help determine whether Musk’s tunnel concept can grow from a relatively small Las Vegas network into a transport system that cities around the world are willing to adopt.<h2>But Boring Company has faced problems</h2>The expansion has not been without controversy. Nevada regulators accused The Boring Company in 2025 of hundreds of environmental violations connected with its Las Vegas construction activities. The company disputed the allegations. Several other projects pitched over the years, including ideas for Baltimore, Chicago and Los Angeles, have also failed to develop into operating networks.That makes the reported $20 billion valuation particularly notable. Investors would effectively be betting that The Boring Company can turn its tunnelling technology and Loop concept into a much larger global infrastructure business. The reported fundraising has not been finalised. But if it closes near the proposed valuation, Musk’s underground transport experiment will have become nearly four times as valuable as it was just four years ago.

Elon Musk’s Boring Company is reportedly seeking around $4 billion at a $20 billion valuation, up sharply from $5.7 billion in 2022. The funding talks come as the tunnelling startup expands its Loop projects from Las Vegas to Nashville and Dubai.


Elon Musk’s underground transportation bet could soon be worth $20 billion. The Boring Company is in talks to raise around $4 billion from investors at a valuation of roughly $20 billion, according to The Wall Street Journal. The deal has not yet closed and its size and terms could still change. If completed at that valuation, it would mark a dramatic jump for the tunnelling startup. The company was valued at about $5.7 billion in 2022, when it raised $675 million from investors including Sequoia Capital, Vy Capital and Founders Fund. What exactly does The Boring Company do? Musk created The Boring Company around a simple but ambitious idea: instead of endlessly expanding congested roads, cities could move some traffic underground. Its best-known operating system is in Las Vegas, where Tesla vehicles transport passengers through tunnels and surface routes connecting stations around the Las Vegas Convention Center and Strip. But the company’s ambitions stretch much further. It is developing the Music City Loop in Nashville, designed to connect downtown with Nashville International Airport. The company says the journey could take around nine minutes. From Las Vegas to Dubai Dubai is another major test. The Boring Company has signed an agreement with Dubai’s Roads and Transport Authority for the Dubai Loop. Its pilot system is planned to include 6.4 kilometres of tunnel and four stations. These projects could help determine whether Musk’s tunnel concept can grow from a relatively small Las Vegas network into a transport system that cities around the world are willing to adopt. But Boring Company has faced problems The expansion has not been without controversy. Nevada regulators accused The Boring Company in 2025 of hundreds of environmental violations connected with its Las Vegas construction activities. The company disputed the allegations. Several other projects pitched over the years, including ideas for Baltimore, Chicago and Los Angeles, have also failed to develop into operating networks. That makes the reported $20 billion valuation particularly notable. Investors would effectively be betting that The Boring Company can turn its tunnelling technology and Loop concept into a much larger global infrastructure business. The reported fundraising has not been finalised. But if it closes near the proposed valuation, Musk’s underground transport experiment will have become nearly four times as valuable as it was just four years ago.

Elon Musk’s Boring Company is reportedly seeking around $4 billion at a $20 billion valuation, up sharply from $5.7 billion in 2022. The funding talks come as the tunnelling startup expands its Loop projects from Las Vegas to Nashville and Dubai.

Elon Musk’s Boring Company eyes $20 billion valuation, seeks $4 billion in fresh funding: Report

ChatGPT may not remain something you open on a phone or computer for much longer. OpenAI president Greg Brockman says the company is “very interested” in bringing deeper ChatGPT integration to smart glasses and other wearable devices, potentially putting its AI assistant directly into products people carry or wear throughout the day.Speaking at a media briefing in New York, Brockman said OpenAI believes its AI interface should eventually be “available on all devices”. The idea goes beyond simply putting the existing ChatGPT app on another screen. OpenAI wants an AI assistant that can understand what users want, interact with software and complete tasks for them.<h2>OpenAI wants to build a ‘Jarvis’-like assistant</h2>Brockman compared the broader vision to Jarvis, the fictional AI assistant used by Tony Stark in the Iron Man universe. OpenAI already has some of the technology needed to move in this direction. Its Codex tools can perform computer-based tasks and interact with software. Brockman believes this could eventually change how people use computers altogether. “I think we’re just going to realise this phase of us, like, clicking things and typing things, it was a phase,” he said. Instead, users could increasingly tell an AI what they want done and allow the agent to handle the steps itself.<h2>Why smart glasses could matter for ChatGPT</h2>Wearables could make such an assistant far more useful. Smart glasses equipped with cameras and microphones can potentially give AI visual and audio context about a user's surroundings. Instead of pulling out a phone, a person could simply speak to the assistant. However, deeper integration requires new technology. Wearables can need continuous background access and different APIs from those designed for conventional phone or desktop apps. Brockman said OpenAI's work on such APIs remains at an early stage.<h2>OpenAI is also building its own hardware</h2>The wearable push does not mean OpenAI has abandoned plans for its own device.Brockman said OpenAI is working on a hardware “solution for a problem that we think is neglected”, without revealing exactly what the product will be. The strategy suggests OpenAI could eventually take two routes: build its own AI hardware while also letting other manufacturers integrate ChatGPT more deeply into their devices. The bigger ambition is clear. OpenAI does not want ChatGPT to remain another app users open. It wants AI to become the interface between people and their technology.

OpenAI is exploring deeper ChatGPT integrations for smart glasses and wearables. President Greg Brockman says the company wants its AI interface available across devices as it works towards assistants that could reduce the need for typing, clicking and opening apps.

ChatGPT may not remain something you open on a phone or computer for much longer. OpenAI president Greg Brockman says the company is “very interested” in bringing deeper ChatGPT integration to smart glasses and other wearable devices, potentially putting its AI assistant directly into products people carry or wear throughout the day. Speaking at a media briefing in New York, Brockman said OpenAI believes its AI interface should eventually be “available on all devices”. The idea goes beyond simply putting the existing ChatGPT app on another screen. OpenAI wants an AI assistant that can understand what users want, interact with software and complete tasks for them. OpenAI wants to build a ‘Jarvis’-like assistant Brockman compared the broader vision to Jarvis, the fictional AI assistant used by Tony Stark in the Iron Man universe. OpenAI already has some of the technology needed to move in this direction. Its Codex tools can perform computer-based tasks and interact with software. Brockman believes this could eventually change how people use computers altogether. “I think we’re just going to realise this phase of us, like, clicking things and typing things, it was a phase,” he said. Instead, users could increasingly tell an AI what they want done and allow the agent to handle the steps itself. Why smart glasses could matter for ChatGPT Wearables could make such an assistant far more useful. Smart glasses equipped with cameras and microphones can potentially give AI visual and audio context about a user's surroundings. Instead of pulling out a phone, a person could simply speak to the assistant. However, deeper integration requires new technology. Wearables can need continuous background access and different APIs from those designed for conventional phone or desktop apps. Brockman said OpenAI's work on such APIs remains at an early stage. OpenAI is also building its own hardware The wearable push does not mean OpenAI has abandoned plans for its own device. Brockman said OpenAI is working on a hardware “solution for a problem that we think is neglected”, without revealing exactly what the product will be. The strategy suggests OpenAI could eventually take two routes: build its own AI hardware while also letting other manufacturers integrate ChatGPT more deeply into their devices. The bigger ambition is clear. OpenAI does not want ChatGPT to remain another app users open. It wants AI to become the interface between people and their technology.

OpenAI is exploring deeper ChatGPT integrations for smart glasses and wearables. President Greg Brockman says the company wants its AI interface available across devices as it works towards assistants that could reduce the need for typing, clicking and opening apps.

Forget typing and clicking: OpenAI wants a ‘Jarvis’-like ChatGPT that works across your devices

After the technology integrated with AI, multiple businesses and giant organisations have begun to increasingly deploy it to handle customer interactions, increasing the efficiency of the firm. A debate is also emerging over what truly defines good customer experience: the ability to mimic human conversation, or the ability to consistently solve problems efficiently. The human-like customer experiences from AI at scale come through conversational natural language processing, real-time sentiment analysis, and predictive personalisation engines. These tools automate routine support while empowering human teams. The integration of technological progress and heightened customer expectations has created an ideal opportunity for revolutionising customer engagement. The next dimension of the customer experience, along with the next best experience approach, is generally led by front-runners who can effectively use AI and gen AI to enhance predictive capabilities and gain a competitive edge, unlocking multiple potential. Reacting to the development, Industry experts argue that while AI has made remarkable strides in communicating naturally, true customer satisfaction hinges less on conversational polish and more on speed, accuracy, and the strength of the underlying data systems powering these interactions.<h3>Combine automation</h3>According to experts, the real transformation lies not in AI sounding more human, but in how effectively organisations can combine automation with human judgment for complex, high-stakes situations, a balance that will define the next phase of customer experience strategy. Prasad Rai, CEO, DAAS LABS, said, "The conversation shouldn't be whether AI can sound like a human, but whether it can consistently make people feel heard, and those are completely different goals. Organisations often confuse the ability to conduct a conversation well with providing good customer experience, when in reality, people care far more about speed, accuracy, and resolution than about how their question was answered. AI can certainly provide a human-like experience, but only when it's built on the right data foundation: proper context, high-quality data, and a clear understanding of what the customer is trying to do." He noted that several organisations fall short because their data is disjointed, their working processes were never designed and developed this way, and no amount of conversational polish can compensate for that. "What AI should really be doing is taking care of the mundane, well-structured decisions and helping teams move faster and smarter, so that people are freed up for the complex, sensitive conversations that genuinely require trust, judgment, and empathy. The companies that succeed in customer experience won't be the ones with the most human-like AI, but the ones with the strongest data foundation underneath it, because that's ultimately what makes everything easier," he added. Highlighting the real customer experience after integration of AI, Sri Mookiah, Founder &amp; CEO, LOWCODEMINDS, stated that for a long time, businesses assumed personalised customer service and scalability were mutually exclusive. AI is now disproving that notion. Moving ahead, competitive advantage won't hinge on which company has the largest AI model, but on which one delivers the superior customer experience. He said, "AI can now understand context, analyse customer history, interpret documents, communicate naturally across voice and chat, and resolve many issues without human intervention. What once required thousands of service agents can increasingly be delivered by intelligent AI systems operating 24/7. But businesses should be careful not to confuse conversation with experience." "Customers do not remember how quickly an answer was generated. They remember whether their problem was understood, resolved fairly, and whether they felt valued. Empathy, accountability, and judgment remain uniquely human strengths, particularly in complex or emotionally charged situations. The organisations that succeed will not replace humans with AI. They will redesign customer journeys where AI handles routine interactions at scale while people focus on trust, relationships, and high-value decisions. The future of customer experience is not artificial or human. It is augmented, where AI provides the speed, and humans provide the judgment that earns lasting customer loyalty," Mookiah added.&nbsp;Also read: <a href="https://www.wionews.com/world/takaichi-s-1-7tn-spending-plan-sparks-fears-of-liz-truss-style-shock-how-will-it-impact-japan-china-and-india-1785059302899">Takaichi's £1.7tn spending plan sparks fears of 'Liz Truss-style' shock: How will it impact Japan, China and India?</a> Giving reality about the customer experience, Hitesh Agrawal, Founder &amp; Managing Director, Them Consulting, stated that the real question is not whether AI can make a difference in customer experience. It has already started. He noted that the real question comes when it actually makes things better for people, and when it quietly makes them worse. "In big banks, the pattern is clear. AI systems now take care of a huge chunk of incoming requests on their own, often 40 to 60 per cent, and in some well‑run setups that climbs to around 80 per cent for simple, routine queries. That shift has completely changed the maths on service speed. First responses that used to take hours now land in minutes. Average query handling times have now been cut to a fraction of what they used to be. For simple things like checking the account balance or confirming a credit card limit, this new speed is genuinely impacting customers positively. Where it gets tricky is what happens beyond those routine moments. When you compare pure AI journeys with human agents, the satisfaction scores are now surprisingly close. On paper, the gap is small. But when you ask people what they actually want when they call in the customer care line in distress, most still will say something similar. If the issue is complex or involves money they are worried about, they would rather talk to a person," he said. "The setups that seem to work best accept that reality instead of fighting it. They use AI to clear out the repetitive work, give agents better context, and then make sure the difficult conversations get to a human quickly, with all the history on the screen. That is what “human‑like at scale” really looks like in practice: machines doing the heavy lifting so people can focus on the moments that actually need them," he added. Sukesh Shetty, Founder &amp; CEO at Away.ai, said, "Most of the AI that got built into customer service last year was built to sound more human. Warmer chatbots. Friendlier scripts with a name and a little personality bolted on. None of that fixed what was actually broken, which is why people still don't trust it. Good customer experience runs on two layers: empathy and information. People are naturally good at empathy. It's the information layer, whether you get the right answer without having to ask twice, that decides if an experience is actually good. Go back and read the complaints people write; nobody's upset about tone; they're upset because someone told them the wrong thing, or told them too late, and that's actually the part AI needs to get right, not how kind it sounds doing it. At Away, our AI doesn't try to be warm with you. It just goes and does the boring part, negotiating with our own supplier network for a better fare. And better isn't always cheaper. Sometimes it's a refundable ticket instead of a locked one, or a lower cancellation fee, and we say so, because most companies would rather imply they found you the cheapest seat than admit they found you a different kind of good deal, and that habit is a big part of why people stopped trusting this stuff. I don't think AI was ever going to earn anyone's trust by sounding more like us. It earns it by being something we can actually rely on when we're not the ones checking."

The human-like customer experiences from AI at scale come through conversational natural language processing, real-time sentiment analysis, and predictive personalisation engines. 

After the technology integrated with AI, multiple businesses and giant organisations have begun to increasingly deploy it to handle customer interactions, increasing the efficiency of the firm. A debate is also emerging over what truly defines good customer experience: the ability to mimic human conversation, or the ability to consistently solve problems efficiently. The human-like customer experiences from AI at scale come through conversational natural language processing, real-time sentiment analysis, and predictive personalisation engines. These tools automate routine support while empowering human teams. The integration of technological progress and heightened customer expectations has created an ideal opportunity for revolutionising customer engagement. The next dimension of the customer experience, along with the next best experience approach, is generally led by front-runners who can effectively use AI and gen AI to enhance predictive capabilities and gain a competitive edge, unlocking multiple potential. Reacting to the development, Industry experts argue that while AI has made remarkable strides in communicating naturally, true customer satisfaction hinges less on conversational polish and more on speed, accuracy, and the strength of the underlying data systems powering these interactions. Combine automation According to experts, the real transformation lies not in AI sounding more human, but in how effectively organisations can combine automation with human judgment for complex, high-stakes situations, a balance that will define the next phase of customer experience strategy. Prasad Rai, CEO, DAAS LABS, said, "The conversation shouldn't be whether AI can sound like a human, but whether it can consistently make people feel heard, and those are completely different goals. Organisations often confuse the ability to conduct a conversation well with providing good customer experience, when in reality, people care far more about speed, accuracy, and resolution than about how their question was answered. AI can certainly provide a human-like experience, but only when it's built on the right data foundation: proper context, high-quality data, and a clear understanding of what the customer is trying to do." He noted that several organisations fall short because their data is disjointed, their working processes were never designed and developed this way, and no amount of conversational polish can compensate for that. "What AI should really be doing is taking care of the mundane, well-structured decisions and helping teams move faster and smarter, so that people are freed up for the complex, sensitive conversations that genuinely require trust, judgment, and empathy. The companies that succeed in customer experience won't be the ones with the most human-like AI, but the ones with the strongest data foundation underneath it, because that's ultimately what makes everything easier," he added. Highlighting the real customer experience after integration of AI, Sri Mookiah, Founder & CEO, LOWCODEMINDS, stated that for a long time, businesses assumed personalised customer service and scalability were mutually exclusive. AI is now disproving that notion. Moving ahead, competitive advantage won't hinge on which company has the largest AI model, but on which one delivers the superior customer experience. He said, "AI can now understand context, analyse customer history, interpret documents, communicate naturally across voice and chat, and resolve many issues without human intervention. What once required thousands of service agents can increasingly be delivered by intelligent AI systems operating 24/7. But businesses should be careful not to confuse conversation with experience." "Customers do not remember how quickly an answer was generated. They remember whether their problem was understood, resolved fairly, and whether they felt valued. Empathy, accountability, and judgment remain uniquely human strengths, particularly in complex or emotionally charged situations. The organisations that succeed will not replace humans with AI. They will redesign customer journeys where AI handles routine interactions at scale while people focus on trust, relationships, and high-value decisions. The future of customer experience is not artificial or human. It is augmented, where AI provides the speed, and humans provide the judgment that earns lasting customer loyalty," Mookiah added. Also read: Takaichi's £1.7tn spending plan sparks fears of 'Liz Truss-style' shock: How will it impact Japan, China and India? Giving reality about the customer experience, Hitesh Agrawal, Founder & Managing Director, Them Consulting, stated that the real question is not whether AI can make a difference in customer experience. It has already started. He noted that the real question comes when it actually makes things better for people, and when it quietly makes them worse. "In big banks, the pattern is clear. AI systems now take care of a huge chunk of incoming requests on their own, often 40 to 60 per cent, and in some well‑run setups that climbs to around 80 per cent for simple, routine queries. That shift has completely changed the maths on service speed. First responses that used to take hours now land in minutes. Average query handling times have now been cut to a fraction of what they used to be. For simple things like checking the account balance or confirming a credit card limit, this new speed is genuinely impacting customers positively. Where it gets tricky is what happens beyond those routine moments. When you compare pure AI journeys with human agents, the satisfaction scores are now surprisingly close. On paper, the gap is small. But when you ask people what they actually want when they call in the customer care line in distress, most still will say something similar. If the issue is complex or involves money they are worried about, they would rather talk to a person," he said. "The setups that seem to work best accept that reality instead of fighting it. They use AI to clear out the repetitive work, give agents better context, and then make sure the difficult conversations get to a human quickly, with all the history on the screen. That is what “human‑like at scale” really looks like in practice: machines doing the heavy lifting so people can focus on the moments that actually need them," he added. Sukesh Shetty, Founder & CEO at Away.ai, said, "Most of the AI that got built into customer service last year was built to sound more human. Warmer chatbots. Friendlier scripts with a name and a little personality bolted on. None of that fixed what was actually broken, which is why people still don't trust it. Good customer experience runs on two layers: empathy and information. People are naturally good at empathy. It's the information layer, whether you get the right answer without having to ask twice, that decides if an experience is actually good. Go back and read the complaints people write; nobody's upset about tone; they're upset because someone told them the wrong thing, or told them too late, and that's actually the part AI needs to get right, not how kind it sounds doing it. At Away, our AI doesn't try to be warm with you. It just goes and does the boring part, negotiating with our own supplier network for a better fare. And better isn't always cheaper. Sometimes it's a refundable ticket instead of a locked one, or a lower cancellation fee, and we say so, because most companies would rather imply they found you the cheapest seat than admit they found you a different kind of good deal, and that habit is a big part of why people stopped trusting this stuff. I don't think AI was ever going to earn anyone's trust by sounding more like us. It earns it by being something we can actually rely on when we're not the ones checking."

The human-like customer experiences from AI at scale come through conversational natural language processing, real-time sentiment analysis, and predictive personalisation engines.&nbsp;

Can AI deliver human-like customer experience at scale? Experts weigh in

A three-month-old AI startup wants to make repetitive office work something humans no longer have to do and investors could value the bet at $1 billion. Prentis, founded by serial entrepreneur Ritankar Das alongside LinkedIn co-founder Reid Hoffman and Zynga founder Mark Pincus, is in talks to raise $100 million at a $1 billion valuation, according to TechCrunch. Launched in April, Prentis is building “computer-use” AI models. Instead of simply answering questions like a chatbot, these models are designed to understand what is happening on a computer screen and operate software directly. The goal is ambitious: teach AI how people actually work, then let agents perform those workflows themselves.<h2>Prentis wants AI to learn your office work</h2>Many workplace processes are never properly documented. Employees simply learn which file to open, which button to click and what to do when something goes wrong. Prentis wants its models to learn these patterns by observing and interacting with real workflows. That could eventually allow customised AI agents to handle tasks such as insurance claims or customs duty refund exceptions without employees manually searching through documents and different systems. The startup has reportedly already signed contracts worth up to $50 million with customers across healthcare and manufacturing.<h2>Prentis claims its smaller AI can beat bigger models&nbsp;</h2> At the centre of the company’s pitch is Hive-32B, its computer-use model. Prentis claims Hive-32B outperforms OpenAI’s GPT-5.4 and Anthropic’s Claude Opus 4.6 on two computer-use benchmarks, WindowsAgentArena and ScreenSpot-v2. More importantly for businesses, Prentis reportedly claims its model can perform tasks at roughly one-tenth the cost of frontier AI APIs. However, these performance claims have not been independently verified.<h2>Why office work could become AI's next battleground</h2>Prentis is betting that computer automation could eventually become a bigger AI use case than coding.But it is entering an increasingly competitive market.OpenAI and Anthropic are also developing AI systems capable of interacting with computers. The attraction is clear: an AI agent that can actually operate workplace software could do far more than simply draft an email or summarise a document. Prentis argues that its advantage comes from training AI on something frontier models may lack exposure to the undocumented steps, exceptions and hand-offs involved in real work. Its small team already includes researchers with experience at OpenAI, Google DeepMind, Meta, Tencent and Alibaba. For Prentis, the bigger bet is that the next stage of the AI race will not simply be about building smarter chatbots. It will be about building AI that can sit at a computer and actually get the work done.

Prentis, founded by Ritankar Das, Reid Hoffman and Mark Pincus, is reportedly seeking $100 million at a $1 billion valuation. Its big bet: AI agents that learn how employees use computers and automate everyday office work.


A three-month-old AI startup wants to make repetitive office work something humans no longer have to do and investors could value the bet at $1 billion. Prentis, founded by serial entrepreneur Ritankar Das alongside LinkedIn co-founder Reid Hoffman and Zynga founder Mark Pincus, is in talks to raise $100 million at a $1 billion valuation, according to TechCrunch. Launched in April, Prentis is building “computer-use” AI models. Instead of simply answering questions like a chatbot, these models are designed to understand what is happening on a computer screen and operate software directly. The goal is ambitious: teach AI how people actually work, then let agents perform those workflows themselves. Prentis wants AI to learn your office work Many workplace processes are never properly documented. Employees simply learn which file to open, which button to click and what to do when something goes wrong. Prentis wants its models to learn these patterns by observing and interacting with real workflows. That could eventually allow customised AI agents to handle tasks such as insurance claims or customs duty refund exceptions without employees manually searching through documents and different systems. The startup has reportedly already signed contracts worth up to $50 million with customers across healthcare and manufacturing. Prentis claims its smaller AI can beat bigger models At the centre of the company’s pitch is Hive-32B, its computer-use model. Prentis claims Hive-32B outperforms OpenAI’s GPT-5.4 and Anthropic’s Claude Opus 4.6 on two computer-use benchmarks, WindowsAgentArena and ScreenSpot-v2. More importantly for businesses, Prentis reportedly claims its model can perform tasks at roughly one-tenth the cost of frontier AI APIs. However, these performance claims have not been independently verified. Why office work could become AI's next battleground Prentis is betting that computer automation could eventually become a bigger AI use case than coding. But it is entering an increasingly competitive market. OpenAI and Anthropic are also developing AI systems capable of interacting with computers. The attraction is clear: an AI agent that can actually operate workplace software could do far more than simply draft an email or summarise a document. Prentis argues that its advantage comes from training AI on something frontier models may lack exposure to the undocumented steps, exceptions and hand-offs involved in real work. Its small team already includes researchers with experience at OpenAI, Google DeepMind, Meta, Tencent and Alibaba. For Prentis, the bigger bet is that the next stage of the AI race will not simply be about building smarter chatbots. It will be about building AI that can sit at a computer and actually get the work done.

Prentis, founded by Ritankar Das, Reid Hoffman and Mark Pincus, is reportedly seeking $100 million at a $1 billion valuation. Its big bet: AI agents that learn how employees use computers and automate everyday office work.

Three-month-old AI startup Prentis, backed by Reid Hoffman, eyes $1 billion valuation with a bet on automating office work

Elon Musk has rejected allegations that he is racist, pointing to his half-Indian partner Shivon Zilis, their four children and the diversity of senior executives working across his companies.During an interview with The Economist, the Tesla and SpaceX chief was confronted directly with the criticism: “People say that you’re a racist.” Musk responded by bringing up his family. “My partner is half-Indian. I have four children with her; one of them was named after a famous Indian physicist,” he said. “So I would say I’m not racist,” Musk added.

<blockquote class="twitter-tweet" data-media-max-width="560">Q: People say that you&#39;re a racist. Elon Musk: My partner is half Indian, and I have four children with her. One of them was named after a famous Indian physicist. So I would say I&#39;m not racist. And also, if you look at the people that are employed at my companies, we have… <a href="https://t.co/w8sfy3F3R7">pic.twitter.com/w8sfy3F3R7</a>&mdash; Clash Report (@clashreport) <a href="https://x.com/clashreport/status/2080339432305996222?ref_src=twsrc%5Etfw">July 23, 2026</a></blockquote> <script async src="https://platform.x.com/widgets.js" charset="utf-8"></script>

<h2>Who is Elon Musk’s half-Indian partner?</h2>Musk was referring to Shivon Zilis, a Canadian-born technology executive with Indian heritage.Zilis joined Musk’s brain-computer interface company Neuralink in 2017 and has served in a senior role at the company. The couple have four children together. One of their sons carries the middle name Sekhar, in honour of Indian-American astrophysicist Subrahmanyan Chandrasekhar. Chandrasekhar won the 1983 Nobel Prize in Physics for his groundbreaking work on the structure and evolution of stars.<h2>Musk points to diversity at his companies</h2>Musk also cited the workforce across his businesses while rejecting the racism allegations. “If you look at people that are employed at my companies, we have senior executives of all races,” Musk said. “I don’t think there is any racism there.” The billionaire was also questioned about whether he was anti-Muslim. Musk rejected that characterisation, saying his objection was to people entering a country with views or behaviour that conflict with its laws and accepted norms.<h2>Musk says he ‘got carried away’ with DOGE</h2>The wide-ranging interview also turned to Musk’s involvement with the Trump administration and the Department of Government Efficiency, or DOGE. Musk admitted he would approach that period differently if given another chance. “I got carried away, frankly,” he said, adding: “I think instead of doing DOGE, I would have basically worked on my companies.” His latest comments come as Musk continues to face scrutiny over his political interventions and statements on race, immigration and culture. The billionaire has also recently drawn controversy over Christopher Nolan’s The Odyssey, criticising the casting of Oscar-winning actress Lupita Nyong’o as Helen of Troy. But when confronted directly over allegations of racism, Musk’s response was clear: he rejected the label and pointed to both his multicultural family and the leadership of his companies.&nbsp;

Elon Musk has rejected racism allegations, citing his half-Indian partner Shivon Zilis, their four children and his son’s middle name honouring Nobel-winning Indian-American astrophysicist Subrahmanyan Chandrasekhar.


Elon Musk has rejected allegations that he is racist, pointing to his half-Indian partner Shivon Zilis, their four children and the diversity of senior executives working across his companies. During an interview with The Economist, the Tesla and SpaceX chief was confronted directly with the criticism: “People say that you’re a racist.” Musk responded by bringing up his family. “My partner is half-Indian. I have four children with her; one of them was named after a famous Indian physicist,” he said. “So I would say I’m not racist,” Musk added.

Elon Musk has rejected racism allegations, citing his half-Indian partner Shivon Zilis, their four children and his son’s middle name honouring Nobel-winning Indian-American astrophysicist Subrahmanyan Chandrasekhar.

‘I have four children with her’: Elon Musk cites half-Indian partner while denying racism claims

Anthropic has launched Claude Opus 5, promising users near-frontier AI intelligence without paying the company's highest price. Released on July 24, Opus 5 is designed for everyday professional work, including software development, research and complex tasks. Anthropic says the model comes close to the intelligence of its more powerful Claude Fable 5 at half the price.The launch also shows where the AI race is heading. Companies are no longer competing only to build the smartest model. They increasingly need models that can deliver strong performance without making every task prohibitively expensive.

<blockquote class="twitter-tweet">Introducing Claude Opus 5. It&#39;s a thoughtful and proactive model that comes close to the frontier intelligence of Fable 5 at half the price. <a href="https://t.co/GQWhcq2CQL">pic.twitter.com/GQWhcq2CQL</a>&mdash; Claude (@claudeai) <a href="https://x.com/claudeai/status/2080699495453528290?ref_src=twsrc%5Etfw">July 24, 2026</a></blockquote> <script async src="https://platform.x.com/widgets.js" charset="utf-8"></script>

<h2>Claude Opus 5: What has changed?</h2>Opus 5 costs the same as its predecessor, Opus 4.8, but Anthropic says it delivers significantly better performance. On coding and knowledge-work evaluations including Frontier-Bench and GDPval-AA, Anthropic says Opus 5 has reached state-of-the-art performance. It is also designed to use computing resources more efficiently, reducing the cost of completing many tasks. The model can adjust how much effort it spends on a problem. Users can therefore favour deeper reasoning for difficult work or use fewer tokens when speed and cost matter more. Opus 5 also has a Fast mode, which Anthropic says operates around 2.5 times faster than its standard mode.<h2>Opus 5 or Fable 5: Which should you use?</h2>Anthropic is not positioning Opus 5 as a replacement for Fable 5. Dianne Penn, an Anthropic product leader, told Reuters that users should choose Opus 5 for value, while Fable 5 remains better suited to highly autonomous projects that could continue for days.That distinction could matter to businesses. Everyday coding and office tasks may not need the company's most expensive frontier model. Anthropic says Opus 5 is now the default model for Claude Max users and the strongest model available to Claude Pro subscribers. It is also available through the Claude API.<h2>Anthropic says Opus 5 is safer, too</h2>There is another notable difference. Opus 5 remains behind Anthropic's Mythos 5 on cybersecurity tasks. Reuters reports that it was less capable of exploiting cyber vulnerabilities than Fable 5, allowing Anthropic to deploy it with less restrictive cyber safeguards. Anthropic also says its automated behavioural testing found Opus 5 to be its most aligned model so far, with lower rates of problematic behaviour. The bigger story, however, is economics. If AI models approaching frontier-level performance can become significantly cheaper, businesses may not always need the most powerful model available. For Anthropic, Opus 5 is its bet that the next stage of the AI race will be fought not just over intelligence, but over how much useful intelligence customers can get for their money.

Anthropic has launched Claude Opus 5, an AI model aimed at coding and everyday professional work. The company says it approaches Fable 5 intelligence at half the price, while offering stronger efficiency and major gains over Opus 4.8.

Anthropic has launched Claude Opus 5, promising users near-frontier AI intelligence without paying the company's highest price. Released on July 24, Opus 5 is designed for everyday professional work, including software development, research and complex tasks. Anthropic says the model comes close to the intelligence of its more powerful Claude Fable 5 at half the price. The launch also shows where the AI race is heading. Companies are no longer competing only to build the smartest model. They increasingly need models that can deliver strong performance without making every task prohibitively expensive.

Anthropic has launched Claude Opus 5, an AI model aimed at coding and everyday professional work. The company says it approaches Fable 5 intelligence at half the price, while offering stronger efficiency and major gains over Opus 4.8.

Anthropic launches Claude Opus 5 with near-Fable 5 intelligence at half the price

Google has discontinued Google Play services for devices running Android 6.0 Marshmallow, bumping the minimum required version up to Android 7.0 Nougat. Devices stuck on the older OS won't stop working immediately. Still, they may gradually lose the ability to download new apps or get Play Store updates through as developers phase out support for older Android versions. The change impacts phones that cannot be updated past Android 6.0, an operating system which came in 2015. Google has updated the minimum Play Services requirement now set at Android 7.0 (API level 24), effectively closing the book on Marshmallow support after years of backing it. Functioning as a background service, Google Play Services underpins a range of core Android features, from signing into Google accounts and receiving push notifications to location tracking, Maps functionality, security verification, and data syncing. It also gives developers the APIs needed to integrate Google-powered capabilities into their apps. Meanwhile, Google Play Services is different from the Google Play Store. Several Android apps depend on it to install, update and function properly. After the end of Play Services support, developers are likely to slowly drop compatibility with Android 6.0 devices, resulting in users eventually discovering certain apps that can no longer be downloaded from the Play Store; however, others might stop receiving updates or fail to work correctly.&nbsp;Also read: <a href="https://www.wionews.com/technology/nvidia-is-digging-its-own-grave-deepseek-ceo-backs-china-s-ai-chip-push-1784913193475">'Nvidia is digging its own grave': DeepSeek CEO backs China's AI chip push</a><h3>Will your phone stop working?</h3>Not entirely. Devices running Android 6.0 Marshmallow will still be able to handle basic functions like calling, texting, viewing photos, and running apps already installed on them. That said, the overall experience will likely worsen with time. Since newer app versions demand updated Play Services and more recent Android builds, users can expect growing compatibility problems. Core functions like signing into Google accounts, using location-based features, and receiving push notifications could also start acting unreliably. Additionally, without further Play Services updates, these phones won't get future security patches, leaving them increasingly exposed to newly emerging threats.

Functioning as a background service, Google Play Services underpins a range of core Android features, from signing into Google accounts and receiving push notifications to location tracking, Maps functionality, security verification, and data syncing. 

Google has discontinued Google Play services for devices running Android 6.0 Marshmallow, bumping the minimum required version up to Android 7.0 Nougat. Devices stuck on the older OS won't stop working immediately. Still, they may gradually lose the ability to download new apps or get Play Store updates through as developers phase out support for older Android versions. The change impacts phones that cannot be updated past Android 6.0, an operating system which came in 2015. Google has updated the minimum Play Services requirement now set at Android 7.0 (API level 24), effectively closing the book on Marshmallow support after years of backing it. Functioning as a background service, Google Play Services underpins a range of core Android features, from signing into Google accounts and receiving push notifications to location tracking, Maps functionality, security verification, and data syncing. It also gives developers the APIs needed to integrate Google-powered capabilities into their apps. Meanwhile, Google Play Services is different from the Google Play Store. Several Android apps depend on it to install, update and function properly. After the end of Play Services support, developers are likely to slowly drop compatibility with Android 6.0 devices, resulting in users eventually discovering certain apps that can no longer be downloaded from the Play Store; however, others might stop receiving updates or fail to work correctly. Also read: 'Nvidia is digging its own grave': DeepSeek CEO backs China's AI chip push Will your phone stop working? Not entirely. Devices running Android 6.0 Marshmallow will still be able to handle basic functions like calling, texting, viewing photos, and running apps already installed on them. That said, the overall experience will likely worsen with time. Since newer app versions demand updated Play Services and more recent Android builds, users can expect growing compatibility problems. Core functions like signing into Google accounts, using location-based features, and receiving push notifications could also start acting unreliably. Additionally, without further Play Services updates, these phones won't get future security patches, leaving them increasingly exposed to newly emerging threats.

Functioning as a background service, Google Play Services underpins a range of core Android features, from signing into Google accounts and receiving push notifications to location tracking, Maps functionality, security verification, and data syncing.&nbsp;

Google ends Play Services Support, likely to stop these Android phones from downloading apps & games

DeepSeek founder Liang Wenfeng has laid out an unusually ambitious vision for the Chinese AI company: focus on artificial general intelligence (AGI), keep commercialisation secondary and help build an AI ecosystem less dependent on Nvidia. His remarks emerged from a lengthy discussion with investors and offer a rare look at how the normally low-profile entrepreneur thinks about the global AI race. While companies across the industry are spending heavily to turn AI into profitable products, Liang said DeepSeek's main objective remains AGI broadly referring to AI capable of performing intellectual tasks across many areas at or beyond human level.<h2>DeepSeek wants the 'watermelon', not the 'sesame seeds'</h2>Liang compared near-term revenue opportunities with “sesame seeds”, suggesting DeepSeek will pick some up without allowing them to distract the company from a much bigger prize. “Our goal should be AGI,” he said. The philosophy does not mean DeepSeek will ignore commercial opportunities. Rather, Liang argues that building more advanced intelligence should take priority over maximising revenue from today's products. He described AI development as a staircase. Language models provide the foundation for reasoning, reasoning enables more capable agents, and future systems could eventually learn and improve continuously. If an AI system can help develop its own successor, Liang believes progress could accelerate dramatically.<h2>Why Liang thinks Nvidia is 'digging its own grave'</h2>Perhaps his most provocative remarks concerned Nvidia. US restrictions have limited Chinese companies' access to some of Nvidia's most advanced AI hardware. But Liang argues that those restrictions are creating a powerful incentive for China to develop alternatives. “If Nvidia cards can't be bought, everyone is forced into it everyone has to do domestic chips,” he said. Liang went further, saying he believes “Nvidia is digging its own grave” by creating conditions that encourage Chinese companies to reduce their dependence on its hardware. DeepSeek is working closely with Huawei and expects to participate deeply in its AI computing ecosystem.<h2>Can Huawei really replace Nvidia?</h2>Liang's comments should not be mistaken for a claim that China has already eliminated Nvidia's technological advantage. He acknowledged a substantial efficiency gap, saying roughly four Huawei chips may be needed for the work of one comparable Nvidia chip, while China's hardware remains about two years behind. His broader argument is about direction rather than today's outright performance. DeepSeek has already demonstrated an interest in designing its software around hardware limitations. Its researchers have previously detailed how DeepSeek-V3 was trained using 2,048 Nvidia H800 GPUs while relying on several hardware-aware efficiency techniques. Liang believes the same pressure could ultimately produce a broader Chinese AI ecosystem built around domestic chips, open models and lower prices. That would turn Washington's restrictions into something of a long-term gamble: limiting China's access to America's best AI hardware today could also give Chinese companies a powerful reason to make sure they need less of it tomorrow.

DeepSeek founder Liang Wenfeng says AGI, not short-term profit, remains the company's main goal. He also argues that US restrictions on Nvidia chips are accelerating China's push towards domestic AI hardware, with DeepSeek working closely with Huawei.


DeepSeek founder Liang Wenfeng has laid out an unusually ambitious vision for the Chinese AI company: focus on artificial general intelligence (AGI), keep commercialisation secondary and help build an AI ecosystem less dependent on Nvidia. His remarks emerged from a lengthy discussion with investors and offer a rare look at how the normally low-profile entrepreneur thinks about the global AI race. While companies across the industry are spending heavily to turn AI into profitable products, Liang said DeepSeek's main objective remains AGI broadly referring to AI capable of performing intellectual tasks across many areas at or beyond human level. DeepSeek wants the 'watermelon', not the 'sesame seeds' Liang compared near-term revenue opportunities with “sesame seeds”, suggesting DeepSeek will pick some up without allowing them to distract the company from a much bigger prize. “Our goal should be AGI,” he said. The philosophy does not mean DeepSeek will ignore commercial opportunities. Rather, Liang argues that building more advanced intelligence should take priority over maximising revenue from today's products. He described AI development as a staircase. Language models provide the foundation for reasoning, reasoning enables more capable agents, and future systems could eventually learn and improve continuously. If an AI system can help develop its own successor, Liang believes progress could accelerate dramatically. Why Liang thinks Nvidia is 'digging its own grave' Perhaps his most provocative remarks concerned Nvidia. US restrictions have limited Chinese companies' access to some of Nvidia's most advanced AI hardware. But Liang argues that those restrictions are creating a powerful incentive for China to develop alternatives. “If Nvidia cards can't be bought, everyone is forced into it everyone has to do domestic chips,” he said. Liang went further, saying he believes “Nvidia is digging its own grave” by creating conditions that encourage Chinese companies to reduce their dependence on its hardware. DeepSeek is working closely with Huawei and expects to participate deeply in its AI computing ecosystem. Can Huawei really replace Nvidia? Liang's comments should not be mistaken for a claim that China has already eliminated Nvidia's technological advantage. He acknowledged a substantial efficiency gap, saying roughly four Huawei chips may be needed for the work of one comparable Nvidia chip, while China's hardware remains about two years behind. His broader argument is about direction rather than today's outright performance. DeepSeek has already demonstrated an interest in designing its software around hardware limitations. Its researchers have previously detailed how DeepSeek-V3 was trained using 2,048 Nvidia H800 GPUs while relying on several hardware-aware efficiency techniques. Liang believes the same pressure could ultimately produce a broader Chinese AI ecosystem built around domestic chips, open models and lower prices. That would turn Washington's restrictions into something of a long-term gamble: limiting China's access to America's best AI hardware today could also give Chinese companies a powerful reason to make sure they need less of it tomorrow.

DeepSeek founder Liang Wenfeng says AGI, not short-term profit, remains the company's main goal. He also argues that US restrictions on Nvidia chips are accelerating China's push towards domestic AI hardware, with DeepSeek working closely with Huawei.

'Nvidia is digging its own grave': DeepSeek CEO backs China's AI chip push

Nvidia CEO Jensen Huang has delivered a striking defence of Taiwan's semiconductor industry, saying the island “saved” America's chip and computing industries rather than taking them away. “Taiwan saved the American semiconductor industry, saved the American computing industry, not the opposite,” Huang told reporters during the opening of Taiwanese manufacturer Wistron's first US manufacturing facility in Fort Worth, Texas. The comments carry particular weight because Huang sits at the centre of the global AI boom. Nvidia designs the GPUs powering many of today's most advanced AI systems, while Taiwan remains a critical part of the supply chain that makes those systems possible.Jensen Huang: Without Taiwan, Japan could have taken the leadHuang argued that the relationship between Taiwan and America's technology industry should be viewed as a partnership rather than a story of one country taking manufacturing from another. Without Taiwan's rise as a technology manufacturing powerhouse, he suggested, Japan could have taken a much larger share of America's semiconductor industry. Huang also said Taiwanese companies expanding manufacturing in America are doing so because it makes business sense. “Taiwan is doing it voluntarily because of its own wisdom, not because it's held a gun to its head,” Huang said. He described the relationship between American and Taiwanese technology companies as one built over decades of cooperation and trust.<h2>A sharp contrast with Trump's 'stolen' chip claim</h2>Huang did not name US President Donald Trump. But his comments offer a sharply different interpretation of how Taiwan became a semiconductor powerhouse. Trump has repeatedly accused Taiwan of taking America's semiconductor business and argued that previous US governments should have used tariffs to prevent chip manufacturing from shifting overseas. Taiwan has rejected that characterisation, arguing that its semiconductor industry was built through decades of investment, engineering expertise and the development of a highly specialised manufacturing ecosystem. Huang's argument goes further: Taiwan's rise, in his view, helped American technology companies remain competitive.<h2>Nvidia's latest AI superchips are now being made in Texas</h2>The setting for Huang's comments was significant. Wistron's new D1 facility in Fort Worth is a 324,000-square-foot manufacturing plant backed by a $700 million investment. The facility is already mass-producing Nvidia's GB300 Grace Blackwell Ultra Superchip in the United States. It is also expected to manufacture Nvidia's next-generation Vera Rubin Superchip. Wistron says the site has created more than 500 jobs, with plans to expand that number to around 1,000 by the end of 2026. For Nvidia, this represents something bigger than opening another factory.The company is trying to build more of the physical infrastructure behind the AI boom inside America from chips and packaging to servers and complete AI systems.And there is an irony at the centre of that effort. As Washington pushes to bring advanced technology manufacturing back to American soil, some of the companies helping make that possible are precisely the Taiwanese manufacturers that Huang believes America should be thanking rather than blaming.

Nvidia CEO Jensen Huang says Taiwan “saved” America's semiconductor and computing industries, challenging claims that it took US chip manufacturing away. His remarks came as Taiwan's Wistron opened a $700 million Nvidia AI manufacturing facility in Texas.


Nvidia CEO Jensen Huang has delivered a striking defence of Taiwan's semiconductor industry, saying the island “saved” America's chip and computing industries rather than taking them away. “Taiwan saved the American semiconductor industry, saved the American computing industry, not the opposite,” Huang told reporters during the opening of Taiwanese manufacturer Wistron's first US manufacturing facility in Fort Worth, Texas. The comments carry particular weight because Huang sits at the centre of the global AI boom. Nvidia designs the GPUs powering many of today's most advanced AI systems, while Taiwan remains a critical part of the supply chain that makes those systems possible. Jensen Huang: Without Taiwan, Japan could have taken the lead Huang argued that the relationship between Taiwan and America's technology industry should be viewed as a partnership rather than a story of one country taking manufacturing from another. Without Taiwan's rise as a technology manufacturing powerhouse, he suggested, Japan could have taken a much larger share of America's semiconductor industry. Huang also said Taiwanese companies expanding manufacturing in America are doing so because it makes business sense. “Taiwan is doing it voluntarily because of its own wisdom, not because it's held a gun to its head,” Huang said. He described the relationship between American and Taiwanese technology companies as one built over decades of cooperation and trust. A sharp contrast with Trump's 'stolen' chip claim Huang did not name US President Donald Trump. But his comments offer a sharply different interpretation of how Taiwan became a semiconductor powerhouse. Trump has repeatedly accused Taiwan of taking America's semiconductor business and argued that previous US governments should have used tariffs to prevent chip manufacturing from shifting overseas. Taiwan has rejected that characterisation, arguing that its semiconductor industry was built through decades of investment, engineering expertise and the development of a highly specialised manufacturing ecosystem. Huang's argument goes further: Taiwan's rise, in his view, helped American technology companies remain competitive. Nvidia's latest AI superchips are now being made in Texas The setting for Huang's comments was significant. Wistron's new D1 facility in Fort Worth is a 324,000-square-foot manufacturing plant backed by a $700 million investment. The facility is already mass-producing Nvidia's GB300 Grace Blackwell Ultra Superchip in the United States. It is also expected to manufacture Nvidia's next-generation Vera Rubin Superchip. Wistron says the site has created more than 500 jobs, with plans to expand that number to around 1,000 by the end of 2026. For Nvidia, this represents something bigger than opening another factory. The company is trying to build more of the physical infrastructure behind the AI boom inside America from chips and packaging to servers and complete AI systems. And there is an irony at the centre of that effort. As Washington pushes to bring advanced technology manufacturing back to American soil, some of the companies helping make that possible are precisely the Taiwanese manufacturers that Huang believes America should be thanking rather than blaming.

Nvidia CEO Jensen Huang says Taiwan “saved” America's semiconductor and computing industries, challenging claims that it took US chip manufacturing away. His remarks came as Taiwan's Wistron opened a $700 million Nvidia AI manufacturing facility in Texas.

'Not the opposite': Jensen Huang says Taiwan saved US chip industry

Elon Musk has offered one of his most striking AI predictions yet: artificial intelligence could become smarter than the combined intelligence of all humans within about five years. Speaking to The Economist, the Tesla and xAI chief said there may eventually be little that AI cannot do better than people, apart from simply “being human”. More strikingly, when asked whether humans would still control AI in 10 years, Musk said that was unlikely. Yet Musk, who has spent years warning about AI risks, is no longer arguing that development should be stopped. Instead, he believes its rapid advance may now be impossible to halt.'Let's enjoy the ride'Musk said AI and robotics still carry risks, but argued that their most likely outcome is “incredible abundance for all”. His current philosophy, he said, is essentially: “Let’s enjoy the ride.” Musk predicted a future in which increasingly capable AI and robots could dramatically increase productivity and make goods and services far more abundant.But there is a catch. If AI becomes vastly more intelligent than humans, Musk suggested that believing people would continue controlling it could be unrealistic. The intelligence gap could eventually resemble the gap between humans and chimpanzees. That makes the values built into powerful AI systems crucial. Musk argued that developers should try to ensure advanced AI cares about humanity and wants people to prosper.

<h2>Musk wants rival AI companies to check each other</h2>Musk also proposed a more unusual approach to AI safety: let competitors inspect each other's most powerful models before they are released. He suggested leading AI companies should hold regular discussions on safety and security. Rival developers could potentially receive a week or two to examine new frontier models and identify serious risks.“The competitors can keep each other honest,” Musk said.Governments, in his view, should intervene when a serious danger is identified and the company responsible fails to address it. Such cooperation could involve some of Musk's biggest rivals, including OpenAI, Anthropic and Google DeepMind.<h2>Musk says AI race has become unstoppable</h2>Musk also acknowledged an irony in his own AI journey.He helped establish OpenAI in 2015 partly as a counterweight to Google's growing AI power. He now believes those actions helped accelerate the very race he once hoped to make safer.“All roads lead to the acceleration of AI,” he said.Despite his long-running dispute with OpenAI CEO Sam Altman, Musk indicated that personal rivalries should not stand in the way of AI safety cooperation. He also offered unusually warm words for Anthropic CEO Dario Amodei, describing him as a principled person who cares about the world's future. The bigger message from Musk is therefore not simply that superintelligent AI is coming. It is that the race may already be impossible to stop and humanity's challenge is increasingly shifting from preventing it to making sure what comes next remains aligned with human interests.

Elon Musk predicts AI could surpass the combined intelligence of humanity within five years and says humans may no longer control it within a decade. He now wants rival AI labs to inspect each other's frontier models and regularly discuss safety risks.


Elon Musk has offered one of his most striking AI predictions yet: artificial intelligence could become smarter than the combined intelligence of all humans within about five years. Speaking to The Economist, the Tesla and xAI chief said there may eventually be little that AI cannot do better than people, apart from simply “being human”. More strikingly, when asked whether humans would still control AI in 10 years, Musk said that was unlikely. Yet Musk, who has spent years warning about AI risks, is no longer arguing that development should be stopped. Instead, he believes its rapid advance may now be impossible to halt. 'Let's enjoy the ride' Musk said AI and robotics still carry risks, but argued that their most likely outcome is “incredible abundance for all”. His current philosophy, he said, is essentially: “Let’s enjoy the ride.” Musk predicted a future in which increasingly capable AI and robots could dramatically increase productivity and make goods and services far more abundant. But there is a catch. If AI becomes vastly more intelligent than humans, Musk suggested that believing people would continue controlling it could be unrealistic. The intelligence gap could eventually resemble the gap between humans and chimpanzees. That makes the values built into powerful AI systems crucial. Musk argued that developers should try to ensure advanced AI cares about humanity and wants people to prosper.

Elon Musk predicts AI could surpass the combined intelligence of humanity within five years and says humans may no longer control it within a decade. He now wants rival AI labs to inspect each other's frontier models and regularly discuss safety risks.

Elon Musk says AI could surpass all human intelligence in 5 years, warns humans may lose control

AMD is making one of its biggest bets yet on the artificial intelligence boom—and on Anthropic. The chipmaker will invest up to $5 billion in the Claude developer, while Anthropic has agreed to deploy up to 2 gigawatts of AMD Instinct MI450-series GPUs. The first 1 GW deployment is expected to begin in the first half of 2027. The agreement could be worth tens of billions of dollars in hardware sales and comes as AI companies race to secure enormous amounts of computing capacity. For AMD, however, there is another prize: gaining ground on Nvidia, which continues to dominate the market for advanced AI accelerators.<h2>Why Anthropic needs 2 GW of AI chips</h2>Demand for Claude and other generative AI services is pushing Anthropic to build an increasingly diverse computing network. Two gigawatts is an extraordinary amount of computing infrastructure. It is roughly equivalent to the electricity output of two large nuclear power reactors, although actual data-centre power consumption will depend on how the infrastructure is deployed. Anthropic plans to use AMD's Helios rack-scale systems, powered by MI450-series GPUs. According to The Wall Street Journal, some chips could be deployed in Anthropic's own data centres, while additional computing capacity could be leased through cloud and specialised "neocloud" providers. AMD is also reportedly discussing providing financial backing for some future Anthropic data-centre leases.<h2>AMD isn't just selling Anthropic chips</h2>The partnership goes deeper than hardware. AMD and Anthropic engineers will work together to optimise Claude workloads for AMD Instinct GPUs and accelerate development of ROCm, AMD's software platform for AI computing. AMD also plans to use Claude extensively across its own engineering and product development teams. The investment in Anthropic will be tied to deployment milestones rather than being delivered all at once.<h2>Nvidia faces growing pressure</h2>Nvidia remains the company to beat in AI infrastructure, helped by both its GPUs and its deeply established CUDA software ecosystem.But AI companies increasingly want alternatives. AMD has been aggressively pursuing this opportunity. Earlier in 2026, it struck a multi-year agreement with Meta covering up to 6 GW of AMD GPUs. Anthropic now adds another major frontier AI developer to its customer list.Anthropic itself is also spreading its compute requirements across multiple suppliers instead of depending on one company. It already has major infrastructure relationships involving Amazon and Google. That diversification is becoming strategically important as AI models require ever-larger clusters and demand for advanced chips continues to soar. The AMD-Anthropic agreement therefore represents more than another giant chip order. It shows how the AI race is evolving into a competition not only over who builds the smartest model, but also over who can secure enough chips, electricity and data-centre capacity to run it at global scale.

AMD will invest up to $5 billion in Anthropic, while the Claude maker plans to deploy up to 2 GW of AMD Instinct MI450-series GPUs. The massive partnership gives Anthropic more AI compute and strengthens AMD's attempt to challenge Nvidia's dominance.


AMD is making one of its biggest bets yet on the artificial intelligence boom—and on Anthropic. The chipmaker will invest up to $5 billion in the Claude developer, while Anthropic has agreed to deploy up to 2 gigawatts of AMD Instinct MI450-series GPUs. The first 1 GW deployment is expected to begin in the first half of 2027. The agreement could be worth tens of billions of dollars in hardware sales and comes as AI companies race to secure enormous amounts of computing capacity. For AMD, however, there is another prize: gaining ground on Nvidia, which continues to dominate the market for advanced AI accelerators. Why Anthropic needs 2 GW of AI chips Demand for Claude and other generative AI services is pushing Anthropic to build an increasingly diverse computing network. Two gigawatts is an extraordinary amount of computing infrastructure. It is roughly equivalent to the electricity output of two large nuclear power reactors, although actual data-centre power consumption will depend on how the infrastructure is deployed. Anthropic plans to use AMD's Helios rack-scale systems, powered by MI450-series GPUs. According to The Wall Street Journal, some chips could be deployed in Anthropic's own data centres, while additional computing capacity could be leased through cloud and specialised "neocloud" providers. AMD is also reportedly discussing providing financial backing for some future Anthropic data-centre leases. AMD isn't just selling Anthropic chips The partnership goes deeper than hardware. AMD and Anthropic engineers will work together to optimise Claude workloads for AMD Instinct GPUs and accelerate development of ROCm, AMD's software platform for AI computing. AMD also plans to use Claude extensively across its own engineering and product development teams. The investment in Anthropic will be tied to deployment milestones rather than being delivered all at once. Nvidia faces growing pressure Nvidia remains the company to beat in AI infrastructure, helped by both its GPUs and its deeply established CUDA software ecosystem. But AI companies increasingly want alternatives. AMD has been aggressively pursuing this opportunity. Earlier in 2026, it struck a multi-year agreement with Meta covering up to 6 GW of AMD GPUs. Anthropic now adds another major frontier AI developer to its customer list. Anthropic itself is also spreading its compute requirements across multiple suppliers instead of depending on one company. It already has major infrastructure relationships involving Amazon and Google. That diversification is becoming strategically important as AI models require ever-larger clusters and demand for advanced chips continues to soar. The AMD-Anthropic agreement therefore represents more than another giant chip order. It shows how the AI race is evolving into a competition not only over who builds the smartest model, but also over who can secure enough chips, electricity and data-centre capacity to run it at global scale.

AMD will invest up to $5 billion in Anthropic, while the Claude maker plans to deploy up to 2 GW of AMD Instinct MI450-series GPUs. The massive partnership gives Anthropic more AI compute and strengthens AMD's attempt to challenge Nvidia's dominance.

AMD to invest up to $5 billion in Anthropic as Claude maker orders 2 GW of AI chips

Microsoft is changing the way it powers Copilot, and the strategy could have major implications for OpenAI, Anthropic and the economics of enterprise AI. CEO Satya Nadella says Microsoft is increasingly using its own MAI family of AI models across products including GitHub Copilot, Excel and Outlook. Instead of relying on the most powerful and expensive frontier model for every request, Microsoft wants to use the right model for each job. The goal is simple: maintain or improve AI performance while using fewer tokens and reducing the enormous computing cost of serving millions of users.<h2>Microsoft's own AI is already taking on GPT and Claude</h2>Microsoft says early results are promising. Its specialised MAI model running in Excel is performing on par with GPT-5.6 for common tasks while being cheaper to operate. In GitHub Copilot, MAI-Code-1-Flash has recorded an approximately 10% higher code acceptance rate than GPT-5.4 Mini and Claude Haiku 4.5 in Microsoft's testing. It also used around 10% fewer median tokens. Rather than training one enormous model to do everything, Microsoft trains specialised models inside environments based on the products in which they will actually operate. That means an Excel AI learns spreadsheet workflows, while a coding model learns how developers work inside GitHub Copilot.<h2>OpenAI and Anthropic are not disappearing</h2>The shift does not mean Microsoft is abandoning frontier models. Nadella said models from OpenAI and Anthropic will remain part of Microsoft's AI orchestration system. Powerful models can handle the hardest requests, while smaller and cheaper MAI models deal with high-volume everyday work. The model itself, Nadella argues, is only one part of an AI product. Memory, tools, context, user interactions and the surrounding software can be just as important.Microsoft calls its continuous optimisation process "hill climbing." Instead of judging models only through standard benchmarks, Microsoft evaluates whether they successfully complete tasks customers actually care about.<h2>Why Microsoft's strategy matters</h2>AI has introduced something unusual into mainstream software: a significant cost every time a user asks the system to perform work. For products serving hundreds of millions of people, even small reductions in inference costs can therefore become extremely valuable. Microsoft's answer is a multi-model system where expensive frontier intelligence is used only when necessary. The company says it is seeing promising early results in GitHub Copilot, Excel and Outlook, with the approach now expanding towards Copilot Chat and PowerPoint. Microsoft also plans to bring the same framework to businesses through Microsoft Foundry, allowing companies to optimise AI agents around their own data, workflows and evaluations. The bigger shift is clear: the next stage of the AI race may not simply be about who builds the world's smartest model. It may be about who can deliver enough intelligence for each task at the lowest possible cost.

Microsoft is expanding its home-grown MAI models across GitHub Copilot, Excel and Outlook. Satya Nadella says the strategy uses specialised, cheaper models for everyday tasks while keeping powerful OpenAI and Anthropic models available for demanding workloads.

Microsoft is changing the way it powers Copilot, and the strategy could have major implications for OpenAI, Anthropic and the economics of enterprise AI. CEO Satya Nadella says Microsoft is increasingly using its own MAI family of AI models across products including GitHub Copilot, Excel and Outlook. Instead of relying on the most powerful and expensive frontier model for every request, Microsoft wants to use the right model for each job. The goal is simple: maintain or improve AI performance while using fewer tokens and reducing the enormous computing cost of serving millions of users. Microsoft's own AI is already taking on GPT and Claude Microsoft says early results are promising. Its specialised MAI model running in Excel is performing on par with GPT-5.6 for common tasks while being cheaper to operate. In GitHub Copilot, MAI-Code-1-Flash has recorded an approximately 10% higher code acceptance rate than GPT-5.4 Mini and Claude Haiku 4.5 in Microsoft's testing. It also used around 10% fewer median tokens. Rather than training one enormous model to do everything, Microsoft trains specialised models inside environments based on the products in which they will actually operate. That means an Excel AI learns spreadsheet workflows, while a coding model learns how developers work inside GitHub Copilot. OpenAI and Anthropic are not disappearing The shift does not mean Microsoft is abandoning frontier models. Nadella said models from OpenAI and Anthropic will remain part of Microsoft's AI orchestration system. Powerful models can handle the hardest requests, while smaller and cheaper MAI models deal with high-volume everyday work. The model itself, Nadella argues, is only one part of an AI product. Memory, tools, context, user interactions and the surrounding software can be just as important. Microsoft calls its continuous optimisation process "hill climbing." Instead of judging models only through standard benchmarks, Microsoft evaluates whether they successfully complete tasks customers actually care about. Why Microsoft's strategy matters AI has introduced something unusual into mainstream software: a significant cost every time a user asks the system to perform work. For products serving hundreds of millions of people, even small reductions in inference costs can therefore become extremely valuable. Microsoft's answer is a multi-model system where expensive frontier intelligence is used only when necessary. The company says it is seeing promising early results in GitHub Copilot, Excel and Outlook, with the approach now expanding towards Copilot Chat and PowerPoint. Microsoft also plans to bring the same framework to businesses through Microsoft Foundry, allowing companies to optimise AI agents around their own data, workflows and evaluations. The bigger shift is clear: the next stage of the AI race may not simply be about who builds the world's smartest model. It may be about who can deliver enough intelligence for each task at the lowest possible cost.

Microsoft is expanding its home-grown MAI models across GitHub Copilot, Excel and Outlook. Satya Nadella says the strategy uses specialised, cheaper models for everyday tasks while keeping powerful OpenAI and Anthropic models available for demanding workloads.

Microsoft has a new plan for AI: Use expensive frontier models only when needed

WASHINGTON – “My main concern is with lethal autonomous weapons systems. Let us call them what they are: killer robots. Machines selecting and engaging their target and taking a life – without human control and judgement. This is morally repugnant.”So said United Nations Secretary-General António Guterres at a meeting in July that brought together AI leaders from UN member states, international organisations, the private sector, academia, and civil society leaders.Officially named the first Global Dialogue on Artificial Intelligence, it reflected growing global concern that rapidly advancing AI is outrunning the ability of governments to regulate it. While that applies to all aspects of modern life, it is particularly worrying in warfare.Also read | <a href="https://www.wionews.com/technology/india-s-next-terror-threat-may-not-cross-the-border-it-may-enter-through-your-phone-wion-decodes-1784346117443">India's next terror threat may not cross the border. It may enter through your phone | WION Decodes</a>Therefore, Guterres argued, the deployment of LAWS (lethal autonomous weapons systems) “is politically unacceptable. And it must be banned by international law.”International law and international bans have a poor track record, but Guterres deserves praise for finally drawing attention to a subject that has been debated for decades by human rights groups, the military and experts working for the world’s biggest AI companies, most of them based in the United States.One of them, Anthropic, is involved in a high-profile dispute with the US Department of Defence that mirrored the concerns voiced by Guterres’s UN conference. Anthropic, along with Google, OpenAI and xAI, won government contracts last year worth around $200 million to accelerate AI adoption for national security.Unlike the other AI companies, Anthropic insisted that its most sophisticated model must not be used by the government for fully autonomous weapons; i.e. robots, and mass domestic surveillance. &nbsp;To emphasise fears that technology is outrunning human control, Anthropic’s CEO, Dario Amodei, called for a global, coordinated slowdown in artificial intelligence development. AI systems, he warned, are advancing with such speed that they soon could build and improve their own successors, without human control.Other AI-developing companies reacted coolly. The Pentagon reacted with anger. It ordered all federal agencies to stop using Anthropic technology and declared the company a “supply chain risk” – a designation that is now the subject of a battle in court. &nbsp;It is more than a contract dispute – it is a fundamental disagreement over who controls the use of AI in weapons or other aspects of life.When you talk to techies, sooner or later the different types of control over aerial, ground and maritime unmanned vehicles come up. &nbsp;Also read | <a href="https://www.wionews.com/world/what-is-the-tower-of-babel-and-why-did-pope-leo-warn-about-it-while-talking-about-ai-wion-decodes-1779790151646">Pope Leo warns humanity risks repeating ancient mistake: What is the Tower of Babel and what does it have to do with AI?</a>There is “in the loop”, meaning that a human operator directs the autonomous system and decides whether or not to kill the target. There is “on the loop” when a human operator does not directly control the automated system but can override its decision. &nbsp;Finally, “off the loop” indicates that the system operates independently and makes life-or-death decisions.For an illustrative example of an “in the loop” operation, let’s look back to the January 3, 2020 assassination of Major-General Qassem Soleimani, one of Iran’s most powerful military leaders, in a drone strike ordered by US President Donald Trump.Soleimani’s movements had been closely tracked for days, and it became known that he was about to visit Baghdad. On arrival, he was greeted by the leader of a pro-Iranian Iraqi militia who escorted him to two cars waiting outside. As they began pulling away, they were hit by 100-pound Hellfire missiles launched from MQ9 Reaper drones high overhead.The cars exploded into fireballs, incinerating Solimani, his host and eight of their aides.Typical for an “in the loop” operation, the American Air Force officer who pushed the button to launch the missile sat in a stationary “cockpit” at Creech Air Force base near Las Vegas, almost 12,000 kilometres from Baghdad.In the six years since the assassination – or “targeted killing” as the US called it- drones have become a central weapon in the wars between Ukraine and Russia, the US and Iran, Sudan, and most of all Israel against Hamas, a campaign which turned Gaza into a field of ruins and killed more than 70,000 Palestinians.Israel’s war on Gaza showed the limits of “in the loop” operations given the green light by military and political leaders. One of the main purposes of using AI in war is to process huge amounts of data to identify suspected enemies. Left to human intelligence officers, that can be a protracted process.It took weeks to follow Solimani and finally pinpoint him at the door of Baghdad airport. In contrast, Israeli AI programmes, based on cellphone data, facial recognition and locations known to frequent Hamas militants, produced a list of 37,000 targets within days of the October 7, 2023 massacre by Hamas of more than 1,200 Israelis.One user of the system, named Lavender, told the Israeli/Palestinian +972 news outlet that “I would invest 20 seconds for each target and do dozens of them every day. I had zero added-value as a human, apart from being a stamp of approval” (of the AI-generated list). &nbsp;Though you could argue that the Israelis followed “in the loop” procedures, simply pressing a fire button in response to a target provided by a computer, without the time to double-check, does not amount to human control. &nbsp;AI-generated target lists are part of the explanation of the high Palestinian death toll in Gaza. Human rights organisations say most of those killed were civilians, or in military jargon “collateral damage.”It is not often that the chief of an AI company and the chief of the United Nations publicly share worries about how speedily technology spreads. But addressing the UN meeting on Artificial Intelligence this month, Antonio Guterres gave an example.“The Internet took 15 years to reach a billion people,” he said. “AI got there in two. And these systems are no longer tools awaiting instruction – they are writing code, acting online and making choices with less and less human oversight.”Numbers from the two wars the United States fought in Iraq and Afghanistan provide another illustration of the astonishing speed of the development of robotic warfare. The US forces that stormed into Iraq in 2003 had no robots on the ground. There were none in Afghanistan either.Just six years later, the two wars were waged with the help of an estimated 12,000 ground-based robots and 7,000 drones.While there is growing understanding that technology is outpacing both scientific understanding and the ability of governments to adapt, the prospects of universal agreement on a solution are remote.Last November, the UN General Assembly issued a resolution on the risks AI can pose to international humanitarian law and an emerging arms race. &nbsp;The vote was 156 in favour and five against. These five were the United States, Russia, North Korea, Belarus and Israel. There were eight abstentions, including China, which is competing with the US on accelerating AI development.A major problem in pushing ahead, according to a report in the MIT Technology Review, is what the computational neuroscientist Uri Maoz calls “Black Boxes. “He wrote: “Having studied intentions in the human brain for decades and in AI systems more recently, I can attest that state-of –the-art AI systems are essentially Black Boxes. &nbsp;“We know the inputs and outputs, but the artificial “brain” processing them remains opaque. Even their creators cannot fully interpret them or understand how they work.” &nbsp;That will be even more true once the countries that are now using drones – Ukraine has emerged as a leader in its fight against the Russian invasion – begin to use large swarms rather than individual drones. This is why scientists have intensified their study of bees.Like bees, drone swarms operate in large numbers, adapt their behaviour in real time, share tasks, communicate with each other and continue with their tasks even after suffering losses.

As artificial intelligence rapidly transforms modern warfare, the United Nations is calling for a global ban on lethal autonomous weapons systems that can identify and engage targets without meaningful human control, warning that the technology is advancing faster than it can be regulated.

WASHINGTON – “My main concern is with lethal autonomous weapons systems. Let us call them what they are: killer robots. Machines selecting and engaging their target and taking a life – without human control and judgement. This is morally repugnant.” So said United Nations Secretary-General António Guterres at a meeting in July that brought together AI leaders from UN member states, international organisations, the private sector, academia, and civil society leaders. Officially named the first Global Dialogue on Artificial Intelligence, it reflected growing global concern that rapidly advancing AI is outrunning the ability of governments to regulate it. While that applies to all aspects of modern life, it is particularly worrying in warfare. Also read | India's next terror threat may not cross the border. It may enter through your phone | WION Decodes Therefore, Guterres argued, the deployment of LAWS (lethal autonomous weapons systems) “is politically unacceptable. And it must be banned by international law.” International law and international bans have a poor track record, but Guterres deserves praise for finally drawing attention to a subject that has been debated for decades by human rights groups, the military and experts working for the world’s biggest AI companies, most of them based in the United States. One of them, Anthropic, is involved in a high-profile dispute with the US Department of Defence that mirrored the concerns voiced by Guterres’s UN conference. Anthropic, along with Google, OpenAI and xAI, won government contracts last year worth around $200 million to accelerate AI adoption for national security. Unlike the other AI companies, Anthropic insisted that its most sophisticated model must not be used by the government for fully autonomous weapons; i.e. robots, and mass domestic surveillance. To emphasise fears that technology is outrunning human control, Anthropic’s CEO, Dario Amodei, called for a global, coordinated slowdown in artificial intelligence development. AI systems, he warned, are advancing with such speed that they soon could build and improve their own successors, without human control. Other AI-developing companies reacted coolly. The Pentagon reacted with anger. It ordered all federal agencies to stop using Anthropic technology and declared the company a “supply chain risk” – a designation that is now the subject of a battle in court. It is more than a contract dispute – it is a fundamental disagreement over who controls the use of AI in weapons or other aspects of life. When you talk to techies, sooner or later the different types of control over aerial, ground and maritime unmanned vehicles come up. Also read | Pope Leo warns humanity risks repeating ancient mistake: What is the Tower of Babel and what does it have to do with AI? There is “in the loop”, meaning that a human operator directs the autonomous system and decides whether or not to kill the target. There is “on the loop” when a human operator does not directly control the automated system but can override its decision. Finally, “off the loop” indicates that the system operates independently and makes life-or-death decisions. For an illustrative example of an “in the loop” operation, let’s look back to the January 3, 2020 assassination of Major-General Qassem Soleimani, one of Iran’s most powerful military leaders, in a drone strike ordered by US President Donald Trump. Soleimani’s movements had been closely tracked for days, and it became known that he was about to visit Baghdad. On arrival, he was greeted by the leader of a pro-Iranian Iraqi militia who escorted him to two cars waiting outside. As they began pulling away, they were hit by 100-pound Hellfire missiles launched from MQ9 Reaper drones high overhead. The cars exploded into fireballs, incinerating Solimani, his host and eight of their aides. Typical for an “in the loop” operation, the American Air Force officer who pushed the button to launch the missile sat in a stationary “cockpit” at Creech Air Force base near Las Vegas, almost 12,000 kilometres from Baghdad. In the six years since the assassination – or “targeted killing” as the US called it- drones have become a central weapon in the wars between Ukraine and Russia, the US and Iran, Sudan, and most of all Israel against Hamas, a campaign which turned Gaza into a field of ruins and killed more than 70,000 Palestinians. Israel’s war on Gaza showed the limits of “in the loop” operations given the green light by military and political leaders. One of the main purposes of using AI in war is to process huge amounts of data to identify suspected enemies. Left to human intelligence officers, that can be a protracted process. It took weeks to follow Solimani and finally pinpoint him at the door of Baghdad airport. In contrast, Israeli AI programmes, based on cellphone data, facial recognition and locations known to frequent Hamas militants, produced a list of 37,000 targets within days of the October 7, 2023 massacre by Hamas of more than 1,200 Israelis. One user of the system, named Lavender, told the Israeli/Palestinian +972 news outlet that “I would invest 20 seconds for each target and do dozens of them every day. I had zero added-value as a human, apart from being a stamp of approval” (of the AI-generated list). Though you could argue that the Israelis followed “in the loop” procedures, simply pressing a fire button in response to a target provided by a computer, without the time to double-check, does not amount to human control. AI-generated target lists are part of the explanation of the high Palestinian death toll in Gaza. Human rights organisations say most of those killed were civilians, or in military jargon “collateral damage.” It is not often that the chief of an AI company and the chief of the United Nations publicly share worries about how speedily technology spreads. But addressing the UN meeting on Artificial Intelligence this month, Antonio Guterres gave an example. “The Internet took 15 years to reach a billion people,” he said. “AI got there in two. And these systems are no longer tools awaiting instruction – they are writing code, acting online and making choices with less and less human oversight.” Numbers from the two wars the United States fought in Iraq and Afghanistan provide another illustration of the astonishing speed of the development of robotic warfare. The US forces that stormed into Iraq in 2003 had no robots on the ground. There were none in Afghanistan either. Just six years later, the two wars were waged with the help of an estimated 12,000 ground-based robots and 7,000 drones. While there is growing understanding that technology is outpacing both scientific understanding and the ability of governments to adapt, the prospects of universal agreement on a solution are remote. Last November, the UN General Assembly issued a resolution on the risks AI can pose to international humanitarian law and an emerging arms race. The vote was 156 in favour and five against. These five were the United States, Russia, North Korea, Belarus and Israel. There were eight abstentions, including China, which is competing with the US on accelerating AI development. A major problem in pushing ahead, according to a report in the MIT Technology Review, is what the computational neuroscientist Uri Maoz calls “Black Boxes. “ He wrote: “Having studied intentions in the human brain for decades and in AI systems more recently, I can attest that state-of –the-art AI systems are essentially Black Boxes. “We know the inputs and outputs, but the artificial “brain” processing them remains opaque. Even their creators cannot fully interpret them or understand how they work.” That will be even more true once the countries that are now using drones – Ukraine has emerged as a leader in its fight against the Russian invasion – begin to use large swarms rather than individual drones. This is why scientists have intensified their study of bees. Like bees, drone swarms operate in large numbers, adapt their behaviour in real time, share tasks, communicate with each other and continue with their tasks even after suffering losses.

As artificial intelligence rapidly transforms modern warfare, the United Nations is calling for a global ban on lethal autonomous weapons systems that can identify and engage targets without meaningful human control, warning that the technology is advancing faster than it can be regulated.

Urgent UN warnings: time to tame killer robots is running out

Google has built an AI model that can find security vulnerabilities in massive codebases, write working exploits to prove they are real, and generate patches to fix them — all autonomously, all within hours. And it has decided that only governments should be allowed to use it.&nbsp;What Gemini 3.5 Flash Cyber Does&nbsp;Released on July 21 alongside two general-purpose models, Gemini 3.5 Flash Cyber is not a chatbot. It is a specialised system built on top of Google's 3.5 Flash architecture, fine-tuned specifically for cybersecurity work: scanning code for vulnerabilities, validating those vulnerabilities by generating actual exploits in sandbox environments, and producing patches.&nbsp;The model powers CodeMender, Google's AI security agent, which invokes Flash Cyber multiple times across a codebase to analyse different code paths simultaneously and produce consolidated vulnerability reports. In Google's own testing, the system found remote code execution flaws in public APIs and discovered a memory-corruption vulnerability in a sensitive production service — all within approximately two hours.&nbsp;The performance numbers are notable. On Google's internal Chrome vulnerability pipeline, Flash Cyber found 55 unique security issues compared to 47 from the standard 3.5 Flash model and 36 from Anthropic's Claude Opus. On the Big Sleep security evaluation, it significantly outperformed both its parent model and the larger 3.6 Flash.&nbsp;The Exploit Generation Problem&nbsp;What makes Flash Cyber different from existing code-scanning tools is the exploit step. The model does not merely flag potential vulnerabilities — it builds working proof-of-concept exploits to verify they are real. In one documented case, the system generated a fully reliable remote code execution exploit that bypassed standard mitigation techniques.&nbsp;This is the capability that forced Google's hand on access restrictions. A model that finds vulnerabilities is a defensive tool. A model that writes working exploits is, by definition, dual-use — and Google has acknowledged that explicitly.&nbsp;Why Only Governments&nbsp;Google DeepMind is restricting Flash Cyber to governments and trusted partners through a limited-access pilot programme. The stated rationale is to give frontline defenders — national cybersecurity agencies, military networks, critical infrastructure operators — the ability to find and patch vulnerabilities before they are exploited, while keeping the exploit-generation capability out of broader circulation.&nbsp;The restriction makes Flash Cyber the first frontier AI model to be built for offensive security applications and simultaneously locked behind a government access gate. It sits in a category that did not exist a year ago: AI that is too dangerous for the private sector but too useful to withhold from the state.&nbsp;The Three-Model Release&nbsp;Flash Cyber was part of a broader July 21 launch that included Gemini 3.6 Flash, a general-purpose model priced at $1.50 per million input tokens, and Gemini 3.5 Flash-Lite, a lightweight variant for high-volume applications. Google also teased Gemini 4 but provided no timeline.&nbsp;Notably absent from the release was Gemini 3.5 Pro, the large reasoning model the AI community has been waiting for since Google previewed its capabilities earlier this year. Google has not explained the delay, and each new Flash release without an accompanying Pro launch intensifies speculation about what is holding it back.&nbsp;Google already uses Flash Cyber internally across Chrome, Android, Cloud, Ads, and YouTube security teams. The model's transition from internal tool to government product marks a new phase in how AI companies think about their most sensitive capabilities — not as products to sell broadly, but as strategic assets to share selectively with the state.

Google DeepMind released Gemini 3.5 Flash Cyber, an AI that autonomously finds vulnerabilities, writes working exploits, and generates patches. Access is restricted to governments and trusted partners — the first frontier AI model locked behind a government gate.

Google has built an AI model that can find security vulnerabilities in massive codebases, write working exploits to prove they are real, and generate patches to fix them — all autonomously, all within hours. And it has decided that only governments should be allowed to use it. What Gemini 3.5 Flash Cyber Does Released on July 21 alongside two general-purpose models, Gemini 3.5 Flash Cyber is not a chatbot. It is a specialised system built on top of Google's 3.5 Flash architecture, fine-tuned specifically for cybersecurity work: scanning code for vulnerabilities, validating those vulnerabilities by generating actual exploits in sandbox environments, and producing patches. The model powers CodeMender, Google's AI security agent, which invokes Flash Cyber multiple times across a codebase to analyse different code paths simultaneously and produce consolidated vulnerability reports. In Google's own testing, the system found remote code execution flaws in public APIs and discovered a memory-corruption vulnerability in a sensitive production service — all within approximately two hours. The performance numbers are notable. On Google's internal Chrome vulnerability pipeline, Flash Cyber found 55 unique security issues compared to 47 from the standard 3.5 Flash model and 36 from Anthropic's Claude Opus. On the Big Sleep security evaluation, it significantly outperformed both its parent model and the larger 3.6 Flash. The Exploit Generation Problem What makes Flash Cyber different from existing code-scanning tools is the exploit step. The model does not merely flag potential vulnerabilities — it builds working proof-of-concept exploits to verify they are real. In one documented case, the system generated a fully reliable remote code execution exploit that bypassed standard mitigation techniques. This is the capability that forced Google's hand on access restrictions. A model that finds vulnerabilities is a defensive tool. A model that writes working exploits is, by definition, dual-use — and Google has acknowledged that explicitly. Why Only Governments Google DeepMind is restricting Flash Cyber to governments and trusted partners through a limited-access pilot programme. The stated rationale is to give frontline defenders — national cybersecurity agencies, military networks, critical infrastructure operators — the ability to find and patch vulnerabilities before they are exploited, while keeping the exploit-generation capability out of broader circulation. The restriction makes Flash Cyber the first frontier AI model to be built for offensive security applications and simultaneously locked behind a government access gate. It sits in a category that did not exist a year ago: AI that is too dangerous for the private sector but too useful to withhold from the state. The Three-Model Release Flash Cyber was part of a broader July 21 launch that included Gemini 3.6 Flash, a general-purpose model priced at $1.50 per million input tokens, and Gemini 3.5 Flash-Lite, a lightweight variant for high-volume applications. Google also teased Gemini 4 but provided no timeline. Notably absent from the release was Gemini 3.5 Pro, the large reasoning model the AI community has been waiting for since Google previewed its capabilities earlier this year. Google has not explained the delay, and each new Flash release without an accompanying Pro launch intensifies speculation about what is holding it back. Google already uses Flash Cyber internally across Chrome, Android, Cloud, Ads, and YouTube security teams. The model's transition from internal tool to government product marks a new phase in how AI companies think about their most sensitive capabilities — not as products to sell broadly, but as strategic assets to share selectively with the state.

Google DeepMind released Gemini 3.5 Flash Cyber on July 21, a specialised AI model that autonomously discovers security vulnerabilities, generates working exploits to verify them, and writes patches — all within hours. Access is restricted exclusively to governments and trusted partners, making it the first frontier AI model built for offence and locked behind a government gate.

Google built a cyber-warfare AI that only governments can use

US artificial intelligence firm Anthropic has been accusing Moonshot AI of engaging in distillation activities. Now, the US government is threatening sanctions against the Chinese AI company, in yet another indication of the two nations' race for tech dominance. Michael Kratsios, Director of the White House Office of Science and Technology Policy, accused Moonshot of undermining American research. The irony is that Moonshot AI was widely praised for drastically reducing the cost of AI compute through open-source technologies. Its release of the massive 2.8-trillion-parameter Kimi K3 model is raising a few questions: did it reach that milestone by exploiting American AI research through model distillation? Here is what you should know.<h2>What did the White House officials say about Moonshot AI?</h2>In a post on X, Michael Kratsios said, "We have information that Moonshot AI distilled Anthropic's Fable for the development of its K3 model. To do this, they developed a sophisticated internal platform to conduct large-scale distillation against US models, allowing them to quickly switch between multiple methods of access to avoid detection."Kratsios alleged that Moonshot AI had acquired Nvidia GB300-equipped servers, including access to GB300 systems in Thailand, likely to train its AI models. "Large-scale, covert industrial distillation aimed at stealing proprietary US technology and undermining American research is unacceptable," he added.US Treasury Secretary Scott Bessent said he did not support the “theft of US IP” while hinting at possible sanctions and Entity List designations against companies found to be engaging in distillation. "We support open-source AI and the innovation it unlocks. But open source is not open season on American IP. When PRC [People's Republic of China] firms conduct covert, industrial-scale distillation attacks that cross the line into IP theft, sanctions and Entity List designations are on the table," Bessent posted on X.Also read: <a href="https://www.wionews.com/world/anthropic-accuses-alibaba-ai-model-distillation-attack-claude-1782377405436">Anthropic accuses Alibaba of massive AI model extraction campaign using 25,000 fake accounts</a>&nbsp;<h2>What is a distillation attack?</h2>Distillation is the process of training a smaller or less capable AI model using the outputs generated by a more advanced model. While model distillation is a legitimate machine learning technique when conducted with permission or on proprietary models, it becomes controversial when companies use fraudulent accounts, proxy services or other methods to repeatedly query proprietary AI systems without authorisation to replicate their capabilities.According to Anthropic, such industrial-scale distillation enables competitors to acquire advanced AI capabilities much faster and at significantly lower cost, effectively bypassing years of research and development. The company argues that this amounts to intellectual property theft when carried out without permission.<a href="https://www.wionews.com/videos/china-s-moonshot-targets-ipo-within-the-next-six-months-1784569359231/embed">https://www.wionews.com/videos/china-s-moonshot-targets-ipo-within-the-next-six-months-1784569359231/embed</a><h2>Anthropic's allegations against Moonshot AI</h2>In a February blog post, Anthropic claimed that it had "identified industrial-scale campaigns" by Chinese AI firms Moonshot AI, DeepSeek and MiniMax to illicitly extract the capabilities of its Claude models to improve their own AI systems.The company alleged that the Chinese AI labs generated more than 16 million exchanges with Claude through approximately 24,000 fraudulent accounts, violating Anthropic's terms of service and regional access restrictions.Specifically regarding Moonshot AI, Anthropic claimed that around 3.4 million exchanges targeted agentic reasoning and tool use, coding and data analysis, computer-use agent development, and computer vision capabilities.Also read: <a href="https://www.wionews.com/world/us-accuses-china-ai-theft-distillation-trump-beijing-summit-1777012695107">What is distilling? US accuses China of ​industrial-scale AI system theft</a><h2>Kimi K3: Moonshot AI's most advanced model</h2>Kimi K3, the 2.8-trillion-parameter open-weight model from Moonshot AI, was released around 16 July. It has narrowed the gap with leading US AI models and outperformed some rivals on coding benchmarks. It was found to be weaker than Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 models on several advanced evaluations.Anthropic had alleged that Kimi relied on hundreds of fraudulent accounts spanning multiple access pathways to conduct distillation attacks. According to Anthropic, the use of varied account types made the campaign harder to detect, while metadata associated with the activity matched the public profiles of senior Moonshot AI staff.The company further claimed that Moonshot AI adopted a more targeted approach, attempting to extract and reconstruct Claude's internal reasoning traces rather than relying solely on output imitation.<h2>Moonshot AI is growing strongly financially</h2>Despite the controversy surrounding its alleged use of stronger AI models, Moonshot AI has continued to grow rapidly and is reportedly considering a Hong Kong initial public offering by the end of the year.The company reportedly reached around $300 million in annual recurring revenue in June, up from $200 million in April 2026 and $100 million in March, driven by paid subscriptions and application programming interface or API usage for its Kimi models.Moonshot AI has raised up to $4 billion in total funding across multiple investment rounds. It is currently valued at around $31.5 billion and is reportedly aiming for a final pre-IPO funding round at a $50 billion valuation in August.

Chinese AI firm Moonshot faces US scrutiny after Anthropic accused it of using industrial-scale distillation to improve its Kimi K3 model. US officials have warned of sanctions and Entity List actions 



US artificial intelligence firm Anthropic has been accusing Moonshot AI of engaging in distillation activities. Now, the US government is threatening sanctions against the Chinese AI company, in yet another indication of the two nations' race for tech dominance. Michael Kratsios, Director of the White House Office of Science and Technology Policy, accused Moonshot of undermining American research. The irony is that Moonshot AI was widely praised for drastically reducing the cost of AI compute through open-source technologies. Its release of the massive 2.8-trillion-parameter Kimi K3 model is raising a few questions: did it reach that milestone by exploiting American AI research through model distillation? Here is what you should know. What did the White House officials say about Moonshot AI? In a post on X, Michael Kratsios said, "We have information that Moonshot AI distilled Anthropic's Fable for the development of its K3 model. To do this, they developed a sophisticated internal platform to conduct large-scale distillation against US models, allowing them to quickly switch between multiple methods of access to avoid detection." Kratsios alleged that Moonshot AI had acquired Nvidia GB300-equipped servers, including access to GB300 systems in Thailand, likely to train its AI models. "Large-scale, covert industrial distillation aimed at stealing proprietary US technology and undermining American research is unacceptable," he added. US Treasury Secretary Scott Bessent said he did not support the “theft of US IP” while hinting at possible sanctions and Entity List designations against companies found to be engaging in distillation. "We support open-source AI and the innovation it unlocks. But open source is not open season on American IP. When PRC [People's Republic of China] firms conduct covert, industrial-scale distillation attacks that cross the line into IP theft, sanctions and Entity List designations are on the table," Bessent posted on X. Also read: Anthropic accuses Alibaba of massive AI model extraction campaign using 25,000 fake accounts What is a distillation attack? Distillation is the process of training a smaller or less capable AI model using the outputs generated by a more advanced model. While model distillation is a legitimate machine learning technique when conducted with permission or on proprietary models, it becomes controversial when companies use fraudulent accounts, proxy services or other methods to repeatedly query proprietary AI systems without authorisation to replicate their capabilities. According to Anthropic, such industrial-scale distillation enables competitors to acquire advanced AI capabilities much faster and at significantly lower cost, effectively bypassing years of research and development. The company argues that this amounts to intellectual property theft when carried out without permission. https://www.wionews.com/videos/china-s-moonshot-targets-ipo-within-the-next-six-months-1784569359231/embed Anthropic's allegations against Moonshot AI In a February blog post, Anthropic claimed that it had "identified industrial-scale campaigns" by Chinese AI firms Moonshot AI, DeepSeek and MiniMax to illicitly extract the capabilities of its Claude models to improve their own AI systems. The company alleged that the Chinese AI labs generated more than 16 million exchanges with Claude through approximately 24,000 fraudulent accounts, violating Anthropic's terms of service and regional access restrictions. Specifically regarding Moonshot AI, Anthropic claimed that around 3.4 million exchanges targeted agentic reasoning and tool use, coding and data analysis, computer-use agent development, and computer vision capabilities. Also read: What is distilling? US accuses China of industrial-scale AI system theft Kimi K3: Moonshot AI's most advanced model Kimi K3, the 2.8-trillion-parameter open-weight model from Moonshot AI, was released around 16 July. It has narrowed the gap with leading US AI models and outperformed some rivals on coding benchmarks. It was found to be weaker than Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 models on several advanced evaluations. Anthropic had alleged that Kimi relied on hundreds of fraudulent accounts spanning multiple access pathways to conduct distillation attacks. According to Anthropic, the use of varied account types made the campaign harder to detect, while metadata associated with the activity matched the public profiles of senior Moonshot AI staff. The company further claimed that Moonshot AI adopted a more targeted approach, attempting to extract and reconstruct Claude's internal reasoning traces rather than relying solely on output imitation. Moonshot AI is growing strongly financially Despite the controversy surrounding its alleged use of stronger AI models, Moonshot AI has continued to grow rapidly and is reportedly considering a Hong Kong initial public offering by the end of the year. The company reportedly reached around $300 million in annual recurring revenue in June, up from $200 million in April 2026 and $100 million in March, driven by paid subscriptions and application programming interface or API usage for its Kimi models. Moonshot AI has raised up to $4 billion in total funding across multiple investment rounds. It is currently valued at around $31.5 billion and is reportedly aiming for a final pre-IPO funding round at a $50 billion valuation in August.

Chinese AI firm Moonshot faces US scrutiny after Anthropic accused it of using industrial-scale distillation to improve its Kimi K3 model. US officials have warned of sanctions and Entity List actions&nbsp;&nbsp;

Moonshot could  face US sanctions: Govt takes note of Chinese AI firm's distillation attacks against Anthropic to improve its Kimi model

Nvidia has offered its clearest look yet at Vera Rubin, the AI platform it hopes will power the next generation of artificial intelligence. During a technical briefing at its headquarters in Santa Clara, California, the company revealed fresh performance figures for the new chip system ahead of AMD's annual AI event. Nvidia says the Vera Rubin NVL72 platform can process 10 times more AI tokens per watt than its current Grace Blackwell system, while delivering significantly higher memory bandwidth and easier deployment.The announcement highlights a major shift in Nvidia's strategy. The company no longer wants to be seen only as the world's leading GPU maker it also wants to become a supplier of complete AI computing systems.<h2>Nvidia is betting on CPUs too</h2>Graphics Processing Units (GPUs) remain the backbone of AI training, but modern AI agents also rely heavily on Central Processing Units (CPUs) to manage data, networking and software operations.Vera Rubin reflects that change.Each NVL72 rack combines 72 Rubin GPUs with 36 Vera CPUs, creating an integrated platform designed for large AI workloads. Nvidia is also planning to sell the Vera CPU separately, expanding its push into a market traditionally dominated by AMD and Intel."We're on a roadmap to crank out new architectures not just GPUs but CPUs," Nvidia Vice President Ian Buck said during the briefing.<h2>Faster, cooler and easier to install</h2>Nvidia says Vera Rubin introduces several practical improvements for data centres.The company has redesigned the platform to use 100 per cent liquid cooling, reducing energy needed for cooling. It has also dramatically reduced cabling, allowing server racks to be installed or replaced in minutes instead of hours.Another key improvement is memory performance. Nvidia claims Vera Rubin offers nearly three times the memory bandwidth of Grace Blackwell, an important advantage as AI models continue to grow in size. The company also revealed that OpenAI is already testing a Vera Rubin rack as an early customer.<h2>A crucial launch in the AI chip race</h2>Vera Rubin arrives at a critical moment for Nvidia. The company is competing aggressively with AMD, which is expected to showcase its Helios AI platform, while cloud providers including Microsoft, Oracle, Meta and Amazon continue investing billions of dollars in AI infrastructure.Nvidia also faces pressure to execute flawlessly after earlier reports of overheating issues affecting some Blackwell server configurations. According to Nvidia, Vera Rubin remains on schedule for full production in the second half of this year. As AI models become larger and more capable, the competition is no longer just about building the fastest GPU. Companies now need complete AI systems that combine processing power, networking, memory efficiency and energy savings and Nvidia is betting Vera Rubin will keep it ahead in that race.

Nvidia has unveiled new performance details for its Vera Rubin AI platform, claiming up to 10 times better token-per-watt efficiency than Grace Blackwell. The system combines CPUs and GPUs to power advanced AI workloads as competition with AMD intensifies.

Nvidia has offered its clearest look yet at Vera Rubin, the AI platform it hopes will power the next generation of artificial intelligence. During a technical briefing at its headquarters in Santa Clara, California, the company revealed fresh performance figures for the new chip system ahead of AMD's annual AI event. Nvidia says the Vera Rubin NVL72 platform can process 10 times more AI tokens per watt than its current Grace Blackwell system, while delivering significantly higher memory bandwidth and easier deployment. The announcement highlights a major shift in Nvidia's strategy. The company no longer wants to be seen only as the world's leading GPU maker it also wants to become a supplier of complete AI computing systems. Nvidia is betting on CPUs too Graphics Processing Units (GPUs) remain the backbone of AI training, but modern AI agents also rely heavily on Central Processing Units (CPUs) to manage data, networking and software operations. Vera Rubin reflects that change. Each NVL72 rack combines 72 Rubin GPUs with 36 Vera CPUs, creating an integrated platform designed for large AI workloads. Nvidia is also planning to sell the Vera CPU separately, expanding its push into a market traditionally dominated by AMD and Intel. "We're on a roadmap to crank out new architectures not just GPUs but CPUs," Nvidia Vice President Ian Buck said during the briefing. Faster, cooler and easier to install Nvidia says Vera Rubin introduces several practical improvements for data centres. The company has redesigned the platform to use 100 per cent liquid cooling, reducing energy needed for cooling. It has also dramatically reduced cabling, allowing server racks to be installed or replaced in minutes instead of hours. Another key improvement is memory performance. Nvidia claims Vera Rubin offers nearly three times the memory bandwidth of Grace Blackwell, an important advantage as AI models continue to grow in size. The company also revealed that OpenAI is already testing a Vera Rubin rack as an early customer. A crucial launch in the AI chip race Vera Rubin arrives at a critical moment for Nvidia. The company is competing aggressively with AMD, which is expected to showcase its Helios AI platform, while cloud providers including Microsoft, Oracle, Meta and Amazon continue investing billions of dollars in AI infrastructure. Nvidia also faces pressure to execute flawlessly after earlier reports of overheating issues affecting some Blackwell server configurations. According to Nvidia, Vera Rubin remains on schedule for full production in the second half of this year. As AI models become larger and more capable, the competition is no longer just about building the fastest GPU. Companies now need complete AI systems that combine processing power, networking, memory efficiency and energy savings and Nvidia is betting Vera Rubin will keep it ahead in that race.

Nvidia has unveiled new performance details for its Vera Rubin AI platform, claiming up to 10 times better token-per-watt efficiency than Grace Blackwell. The system combines CPUs and GPUs to power advanced AI workloads as competition with AMD intensifies.

Nvidia says its next AI superchip is 10 times more efficient than Blackwell

One of the biggest legal battles over artificial intelligence has reached a major milestone. A federal judge in San Francisco has granted final approval to a $1.5 billion settlement between Anthropic and a class of authors who accused the AI company of illegally downloading copyrighted books from online piracy libraries to build its training collection. The ruling closes what the court described as the largest copyright class action settlement in history. Importantly, however, the lawsuit was not about whether AI can legally learn from copyrighted books. Instead, it focused on how Anthropic obtained those books.<h2>What was Anthropic accused of?</h2>The lawsuit claimed Anthropic downloaded copyrighted books from pirate libraries, including LibGen and PiLiMi, before using them to create its internal book collection Earlier in the case, US District Judge Araceli Martínez-Olguín ruled that training AI models on copyrighted material constituted fair use under US copyright law. That decision has become one of the most closely watched rulings in the AI industry.The settlement resolves only the claims related to Anthropic's acquisition of the books not the legality of AI training itself.<h2>What authors will receive</h2>Under the agreement, eligible authors and publishers whose books appeared on Anthropic's Works List can claim around $3,000 per book, roughly four times the statutory minimum available in many copyright cases. According to court records, more than 440,000 books representing over 91 per cent of eligible works have already been claimed. Anthropic must also permanently delete the pirated digital copies it downloaded.<h2>Why this case matters for AI</h2>The settlement gives Anthropic protection only against claims linked to its past acquisition of the pirated books. Judge Martínez-Olguín made it clear that the agreement does not prevent future lawsuits over how Anthropic's AI models generate content or any future copyright disputes. The court also rejected 54 objections seeking broader remedies, including forcing Anthropic to provide source attribution, expand the list of covered works or even delete its AI models, ruling those issues fell outside the scope of this case. The decision is likely to influence other AI copyright lawsuits involving companies such as OpenAI, Meta and Google, which also face legal scrutiny over how training data is collected and used. For the AI industry, the outcome draws an important legal distinction: how copyrighted material is obtained may create liability, while training AI models on that material can still be considered fair use under existing US court rulings.

A US judge has approved a $1.5 billion settlement between Anthropic and authors over pirated books used to build its training library. The company must compensate rights holders and delete illegal copies, while the court's earlier ruling that AI training is fair use remains unchanged.

One of the biggest legal battles over artificial intelligence has reached a major milestone. A federal judge in San Francisco has granted final approval to a $1.5 billion settlement between Anthropic and a class of authors who accused the AI company of illegally downloading copyrighted books from online piracy libraries to build its training collection. The ruling closes what the court described as the largest copyright class action settlement in history. Importantly, however, the lawsuit was not about whether AI can legally learn from copyrighted books. Instead, it focused on how Anthropic obtained those books. What was Anthropic accused of? The lawsuit claimed Anthropic downloaded copyrighted books from pirate libraries, including LibGen and PiLiMi, before using them to create its internal book collection Earlier in the case, US District Judge Araceli Martínez-Olguín ruled that training AI models on copyrighted material constituted fair use under US copyright law. That decision has become one of the most closely watched rulings in the AI industry. The settlement resolves only the claims related to Anthropic's acquisition of the books not the legality of AI training itself. What authors will receive Under the agreement, eligible authors and publishers whose books appeared on Anthropic's Works List can claim around $3,000 per book, roughly four times the statutory minimum available in many copyright cases. According to court records, more than 440,000 books representing over 91 per cent of eligible works have already been claimed. Anthropic must also permanently delete the pirated digital copies it downloaded. Why this case matters for AI The settlement gives Anthropic protection only against claims linked to its past acquisition of the pirated books. Judge Martínez-Olguín made it clear that the agreement does not prevent future lawsuits over how Anthropic's AI models generate content or any future copyright disputes. The court also rejected 54 objections seeking broader remedies, including forcing Anthropic to provide source attribution, expand the list of covered works or even delete its AI models, ruling those issues fell outside the scope of this case. The decision is likely to influence other AI copyright lawsuits involving companies such as OpenAI, Meta and Google, which also face legal scrutiny over how training data is collected and used. For the AI industry, the outcome draws an important legal distinction: how copyrighted material is obtained may create liability, while training AI models on that material can still be considered fair use under existing US court rulings.

A US judge has approved a $1.5 billion settlement between Anthropic and authors over pirated books used to build its training library. The company must compensate rights holders and delete illegal copies, while the court's earlier ruling that AI training is fair use remains unchanged.

Anthropic's $1.5 billion settlement explained: What authors get and what changes

Buying a new smartphone has already become more expensive, and the trend may continue. According to a Nikkei Asia report, Taiwan Semiconductor Manufacturing Company (TSMC) is planning to increase the prices it charges customers for making chips from 2027. The reported increase of 5 to 10 per cent, depending on the customer and chip type, could eventually affect the prices of products from companies such as Apple, Qualcomm, MediaTek, Google, Nvidia and many others. While TSMC has not confirmed the reported price hike, it said its pricing strategy is "strategic, not opportunistic" and is aimed at reflecting the value it delivers to customers.<h2>Why does this matter for smartphone buyers?</h2>TSMC is the world's largest contract chip manufacturer. It produces processors used in millions of smartphones, tablets, laptops and AI devices every year. If the cost of manufacturing chips rises, smartphone brands will also face higher production costs. Although companies may absorb some of the increase, analysts expect at least part of it to be passed on to consumers through higher retail prices. According to the report, customers placing additional orders for high-performance AI chips beyond their original forecasts could even pay an extra 10 to 15 per cent premium, making advanced processors significantly more expensive.<h2>Why are electronics becoming costlier?</h2>The reported move comes as the consumer electronics industry is already dealing with rising component costs. Over the past year, brands including Samsung, Xiaomi, OnePlus, Nothing and Realme have increased prices across several smartphone models. Apple has also raised prices on some MacBook and iPad models in India, with CEO Tim Cook recently saying higher component costs have made price increases difficult to avoid.Consumers are already feeling the impact. A few years ago, smartphones priced below Rs 30,000 often offered flagship-like displays, fast processors and premium cameras. Today, many devices in the same segment make compromises on performance, cameras or battery capacity to keep prices competitive.<h2>Why is TSMC increasing prices?</h2>According to the report, TSMC needs additional revenue to fund massive investments in advanced semiconductor manufacturing. The company is expanding production overseas, including major facilities in Arizona, while investing billions of dollars in next-generation 2nm chip technology.TSMC Chairman C.C. Wei recently said the company prefers gradual price increases rather than sudden, steep hikes that could hurt customers. If the reported pricing changes take effect in 2027, they could influence not only smartphones but also AI servers, laptops, gaming devices and other products that depend on advanced chips. As artificial intelligence drives demand for more powerful processors, the cost of making those chips is becoming one of the biggest factors shaping the future price of consumer technology.

TSMC is reportedly planning to raise chip manufacturing prices by 5-10 per cent from 2027. Since it supplies chips for Apple, Qualcomm, MediaTek, Google and Nvidia, the move could eventually make smartphones, laptops, tablets and other electronic devices more expensive.

Buying a new smartphone has already become more expensive, and the trend may continue. According to a Nikkei Asia report, Taiwan Semiconductor Manufacturing Company (TSMC) is planning to increase the prices it charges customers for making chips from 2027. The reported increase of 5 to 10 per cent, depending on the customer and chip type, could eventually affect the prices of products from companies such as Apple, Qualcomm, MediaTek, Google, Nvidia and many others. While TSMC has not confirmed the reported price hike, it said its pricing strategy is "strategic, not opportunistic" and is aimed at reflecting the value it delivers to customers. Why does this matter for smartphone buyers? TSMC is the world's largest contract chip manufacturer. It produces processors used in millions of smartphones, tablets, laptops and AI devices every year. If the cost of manufacturing chips rises, smartphone brands will also face higher production costs. Although companies may absorb some of the increase, analysts expect at least part of it to be passed on to consumers through higher retail prices. According to the report, customers placing additional orders for high-performance AI chips beyond their original forecasts could even pay an extra 10 to 15 per cent premium, making advanced processors significantly more expensive. Why are electronics becoming costlier? The reported move comes as the consumer electronics industry is already dealing with rising component costs. Over the past year, brands including Samsung, Xiaomi, OnePlus, Nothing and Realme have increased prices across several smartphone models. Apple has also raised prices on some MacBook and iPad models in India, with CEO Tim Cook recently saying higher component costs have made price increases difficult to avoid. Consumers are already feeling the impact. A few years ago, smartphones priced below Rs 30,000 often offered flagship-like displays, fast processors and premium cameras. Today, many devices in the same segment make compromises on performance, cameras or battery capacity to keep prices competitive. Why is TSMC increasing prices? According to the report, TSMC needs additional revenue to fund massive investments in advanced semiconductor manufacturing. The company is expanding production overseas, including major facilities in Arizona, while investing billions of dollars in next-generation 2nm chip technology. TSMC Chairman C.C. Wei recently said the company prefers gradual price increases rather than sudden, steep hikes that could hurt customers. If the reported pricing changes take effect in 2027, they could influence not only smartphones but also AI servers, laptops, gaming devices and other products that depend on advanced chips. As artificial intelligence drives demand for more powerful processors, the cost of making those chips is becoming one of the biggest factors shaping the future price of consumer technology.

TSMC is reportedly planning to raise chip manufacturing prices by 5-10 per cent from 2027. Since it supplies chips for Apple, Qualcomm, MediaTek, Google and Nvidia, the move could eventually make smartphones, laptops, tablets and other electronic devices more expensive.

TSMC plans 2027 chip price increase, report says: Here's why your next smartphone may become more expensive

In India, conversations about successful engineers often begin with the Indian Institutes of Technology (IITs). But OpenAI's newest technology leader tells a different story. Uday Ruddarraju, who studied Computer Science at Chaitanya Bharathi Institute of Technology (CBIT) in Hyderabad, has been promoted to CTO, Compute at OpenAI. The role puts him in charge of expanding the enormous computing infrastructure that powers some of the world's most advanced artificial intelligence models, including GPT-5.6. Announcing the promotion on LinkedIn, Ruddarraju said he was excited to lead OpenAI's effort to build "the world's largest compute footprint" so frontier AI can reach more people and businesses.<h2>From Hyderabad to Silicon Valley</h2>Unlike many Indian technology leaders, Ruddarraju is not an IIT, NIT or IIIT graduate. After completing his engineering degree at CBIT, he interned at Amazon before moving to the United States to pursue a Master's degree in Computer Science at the University of Minnesota. That decision became the foundation of a career focused on one of technology's most valuable specialisations large-scale cloud infrastructure.<h2>Building the invisible engine behind AI</h2>Ruddarraju is not known for creating AI chatbots or inventing language models. Instead, he builds the enormous computing systems that make them possible. These include GPU clusters, high-speed networking, distributed computing platforms, storage systems and AI training infrastructure capable of handling billions of calculations every second.His career reflects that expertise. He spent more than five years at eBay, helping build cloud infrastructure for one of the world's largest e-commerce platforms. He then joined Robinhood, where he spent nearly six years, eventually becoming Senior Director of Engineering and Head of Infrastructure during a period of rapid growth. In 2024, he was hired by xAI, where he led infrastructure engineering for Elon Musk's AI company before joining OpenAI in 2025 as Head of Compute and Infrastructure. Just over a year later, he has been promoted to CTO.<h2>Why compute has become AI's biggest battleground</h2>The promotion highlights a major shift in the AI industry.Building powerful AI models is no longer just about hiring researchers. It also requires vast computing infrastructure that costs billions of dollars and consumes enormous amounts of energy. Companies such as OpenAI, Google, Microsoft, Meta, Anthropic and xAI are investing heavily in AI supercomputers because access to compute has become one of the industry's biggest competitive advantages.As CTO, Compute, Ruddarraju will oversee OpenAI's efforts to rapidly expand that infrastructure, ensuring future frontier models can be trained faster and deployed at global scale. His appointment also sends a broader message to aspiring engineers: while prestigious institutions can open doors, long-term success in technology is increasingly shaped by specialised skills, execution and the ability to solve complex engineering problems.

OpenAI has promoted Indian-origin engineer Uday Ruddarraju to CTO, Compute. A graduate of Hyderabad's CBIT, he now leads the computing infrastructure behind OpenAI's frontier AI models, showing that specialised expertise not just elite college credentials can shape careers in global technology.

In India, conversations about successful engineers often begin with the Indian Institutes of Technology (IITs). But OpenAI's newest technology leader tells a different story. Uday Ruddarraju, who studied Computer Science at Chaitanya Bharathi Institute of Technology (CBIT) in Hyderabad, has been promoted to CTO, Compute at OpenAI. The role puts him in charge of expanding the enormous computing infrastructure that powers some of the world's most advanced artificial intelligence models, including GPT-5.6. Announcing the promotion on LinkedIn, Ruddarraju said he was excited to lead OpenAI's effort to build "the world's largest compute footprint" so frontier AI can reach more people and businesses. From Hyderabad to Silicon Valley Unlike many Indian technology leaders, Ruddarraju is not an IIT, NIT or IIIT graduate. After completing his engineering degree at CBIT, he interned at Amazon before moving to the United States to pursue a Master's degree in Computer Science at the University of Minnesota. That decision became the foundation of a career focused on one of technology's most valuable specialisations large-scale cloud infrastructure. Building the invisible engine behind AI Ruddarraju is not known for creating AI chatbots or inventing language models. Instead, he builds the enormous computing systems that make them possible. These include GPU clusters, high-speed networking, distributed computing platforms, storage systems and AI training infrastructure capable of handling billions of calculations every second. His career reflects that expertise. He spent more than five years at eBay, helping build cloud infrastructure for one of the world's largest e-commerce platforms. He then joined Robinhood, where he spent nearly six years, eventually becoming Senior Director of Engineering and Head of Infrastructure during a period of rapid growth. In 2024, he was hired by xAI, where he led infrastructure engineering for Elon Musk's AI company before joining OpenAI in 2025 as Head of Compute and Infrastructure. Just over a year later, he has been promoted to CTO. Why compute has become AI's biggest battleground The promotion highlights a major shift in the AI industry. Building powerful AI models is no longer just about hiring researchers. It also requires vast computing infrastructure that costs billions of dollars and consumes enormous amounts of energy. Companies such as OpenAI, Google, Microsoft, Meta, Anthropic and xAI are investing heavily in AI supercomputers because access to compute has become one of the industry's biggest competitive advantages. As CTO, Compute, Ruddarraju will oversee OpenAI's efforts to rapidly expand that infrastructure, ensuring future frontier models can be trained faster and deployed at global scale. His appointment also sends a broader message to aspiring engineers: while prestigious institutions can open doors, long-term success in technology is increasingly shaped by specialised skills, execution and the ability to solve complex engineering problems.

OpenAI has promoted Indian-origin engineer Uday Ruddarraju to CTO, Compute. A graduate of Hyderabad's CBIT, he now leads the computing infrastructure behind OpenAI's frontier AI models, showing that specialised expertise not just elite college credentials can shape careers in global technology.

Who is Uday Ruddarraju? The Indian origin engineer now leading OpenAI's compute future

Amid rising fears of the threat posed by AI, OpenAI has revealed that an autonomous AI agent powered by its advanced models went rogue during a security test and carried out a cyberattack on AI platform Hugging Face. The incident has raised fresh concerns over the risks involved in using advanced artificial intelligence systems.<h2>What happened during the incident?</h2>In a blog post on Tuesday (Jul 21), OpenAI explained an “unprecedented cyber incident” &nbsp;involving advanced AI-driven hacking capabilities. The company said that the incident occurred during an internal cybersecurity evaluation designed to test the limits of its AI models’ cyber abilities. The company said the models were placed in a highly isolated environment but managed to exploit a zero-day vulnerability in a package registry cache proxy to gain internet access.The AI agents then identified and chained multiple vulnerabilities across OpenAI’s research systems and Hugging Face’s infrastructure while attempting to complete their assigned task. OpenAI said the models used stolen credentials and other attack methods to access restricted information from Hugging Face’s production environment. The company said the activity was detected by security teams, while Hugging Face’s security systems also identified and contained the breach.Also read |<a href="https://www.wionews.com/world/missed-the-mark-meta-removes-instagram-ai-image-feature-after-backlash-1783737556704"> ‘Missed the mark’: Meta removes Instagram AI image feature after backlash over privacy</a><h2>What did OpenAI say?</h2>Following the incident, OpenAI said it was strengthening its safeguards and improving security measures. It added that it was tightening infrastructure controls, patching a disclosed zero-day vulnerability, and enhancing AI safety measures. Both companies said that they were investigating the incident.“The primary lesson from this incident is that model security and safety must keep pace with rapidly advancing capabilities. We are strengthening the containment, monitoring, access controls, and evaluation practices used during model development,” OpenAI said.Hugging Face, a popular platform that hosts open-source AI models and datasets, had earlier disclosed that it suffered a sophisticated cyberattack unlike previous incidents. The company said the attack appeared to have been conducted entirely by an autonomous AI agent.Also read | <a href="https://www.wionews.com/india-news/india-orders-meta-to-remove-child-sexual-abuse-ads-from-instagram-demands-explaination-in-7-days-1783235163711">India orders Meta to remove Instagram ads promoting child sexual abuse, demands explanation in 7 days: Report</a>In a statement, Hugging Face Co-founder and CEO Clem Delangue said, “We’re grateful for the collaboration with OpenAI on this and other topics. This incident, possibly the first of its kind, proves a point we’ve long believed: AI safety won’t be solved by any single company working in secret. It will be solved in the open, collaboratively, with broad access to AI for every defender, everywhere.”<h2>Concerns over AI</h2>The disclosure has intensified concerns among cybersecurity experts and policymakers about the ability of frontier AI models to operate independently and potentially cause harm.US Representative Greg Casar called the incident alarming and urged stronger safeguards, including mandatory independent safety testing, reporting of AI-related security breaches and global cooperation on AI regulation.Cybersecurity experts warned that the incident could represent a preview of future threats. Katie Moussouris, CEO of Luta Security, compared advanced AI models to highly adaptive systems capable of finding ways around restrictions.Matt Suiche, an engineer at AI cybersecurity firm Tolmo, said the incident showed that frontier AI models are rapidly approaching the capabilities of sophisticated attackers. He added that similar AI-driven cyber risks are no longer limited to top research labs and can emerge from widely available technologies.Watch | The Internet’s Darkest Secret: Inside the global network exploiting children online<a href="https://www.wionews.com/videos/the-internet-s-darkest-secret-inside-the-global-network-exploiting-children-online-1784689947807/embed">https://www.wionews.com/videos/the-internet-s-darkest-secret-inside-the-global-network-exploiting-children-online-1784689947807/embed</a>&nbsp;

OpenAI said an AI agent went rogue during a security test, escaped containment and hacked Hugging Face’s infrastructure, raising concerns over the cybersecurity risks posed by advanced autonomous AI systems.

Amid rising fears of the threat posed by AI, OpenAI has revealed that an autonomous AI agent powered by its advanced models went rogue during a security test and carried out a cyberattack on AI platform Hugging Face. The incident has raised fresh concerns over the risks involved in using advanced artificial intelligence systems. What happened during the incident? In a blog post on Tuesday (Jul 21), OpenAI explained an “unprecedented cyber incident” involving advanced AI-driven hacking capabilities. The company said that the incident occurred during an internal cybersecurity evaluation designed to test the limits of its AI models’ cyber abilities. The company said the models were placed in a highly isolated environment but managed to exploit a zero-day vulnerability in a package registry cache proxy to gain internet access. The AI agents then identified and chained multiple vulnerabilities across OpenAI’s research systems and Hugging Face’s infrastructure while attempting to complete their assigned task. OpenAI said the models used stolen credentials and other attack methods to access restricted information from Hugging Face’s production environment. The company said the activity was detected by security teams, while Hugging Face’s security systems also identified and contained the breach. Also read | ‘Missed the mark’: Meta removes Instagram AI image feature after backlash over privacy What did OpenAI say? Following the incident, OpenAI said it was strengthening its safeguards and improving security measures. It added that it was tightening infrastructure controls, patching a disclosed zero-day vulnerability, and enhancing AI safety measures. Both companies said that they were investigating the incident. “The primary lesson from this incident is that model security and safety must keep pace with rapidly advancing capabilities. We are strengthening the containment, monitoring, access controls, and evaluation practices used during model development,” OpenAI said. Hugging Face, a popular platform that hosts open-source AI models and datasets, had earlier disclosed that it suffered a sophisticated cyberattack unlike previous incidents. The company said the attack appeared to have been conducted entirely by an autonomous AI agent. Also read | India orders Meta to remove Instagram ads promoting child sexual abuse, demands explanation in 7 days: Report In a statement, Hugging Face Co-founder and CEO Clem Delangue said, “We’re grateful for the collaboration with OpenAI on this and other topics. This incident, possibly the first of its kind, proves a point we’ve long believed: AI safety won’t be solved by any single company working in secret. It will be solved in the open, collaboratively, with broad access to AI for every defender, everywhere.” Concerns over AI The disclosure has intensified concerns among cybersecurity experts and policymakers about the ability of frontier AI models to operate independently and potentially cause harm. US Representative Greg Casar called the incident alarming and urged stronger safeguards, including mandatory independent safety testing, reporting of AI-related security breaches and global cooperation on AI regulation. Cybersecurity experts warned that the incident could represent a preview of future threats. Katie Moussouris, CEO of Luta Security, compared advanced AI models to highly adaptive systems capable of finding ways around restrictions. Matt Suiche, an engineer at AI cybersecurity firm Tolmo, said the incident showed that frontier AI models are rapidly approaching the capabilities of sophisticated attackers. He added that similar AI-driven cyber risks are no longer limited to top research labs and can emerge from widely available technologies. Watch | The Internet’s Darkest Secret: Inside the global network exploiting children online https://www.wionews.com/videos/the-internet-s-darkest-secret-inside-the-global-network-exploiting-children-online-1784689947807/embed

OpenAI said an AI agent went rogue during a security test, escaped containment and hacked Hugging Face’s infrastructure, raising concerns over the cybersecurity risks posed by advanced autonomous AI systems.

OpenAI’s AI agent gone rogue? Company reveals chilling hack at startup

A US court has begun a trial against Meta Platforms, the parent company of Facebook, Instagram and WhatsApp, over allegations that it deliberately designed addictive features targeting teenagers. Jury selection began in Nashville, Tennessee, on Monday (Jul 20), in a lawsuit brought by the Tennessee Attorney General under the state's consumer protection law.<h2>'Instagram, the addiction machine'</h2>Tennessee Attorney General Jonathan Skrmetti accused Meta of knowingly designing Instagram as an "addiction machine" that drives compulsive use among teenagers. The lawsuit alleged that Meta failed to disclose its own internal research on the platform's harms to young users and misled the public about its safety. According to the complaint, Meta continued deploying features it allegedly knew were harmful, while downplaying the associated risks.Meta chief executive Mark Zuckerberg previously testified that the company's algorithms are not intentionally designed to make young users addicted and that Meta has invested in safety features for teenagers.The trial is expected to last six to seven weeks. If Meta is found liable, the court could impose civil penalties of up to US$1,000 for each violation of Tennessee's Consumer Protection Act. it could also order changes to Instagram's design and practices.&nbsp;Also read: <a href="https://www.wionews.com/world/broken-ai-useless-products-mark-zuckerberg-is-steering-meta-towards-a-disaster-1780207649003">'Broken AI, useless products': Mark Zuckerberg might be steering Meta towards a disaster</a><h2>Meta is facing several cases across the US</h2>The Tennessee case overlaps with a growing number of lawsuits against Meta alleging that its platforms were deliberately designed to maximise engagement at the expense of young users' mental health.In May this year, Meta settled a lawsuit brought by Breathitt County Schools in Kentucky for US$9 million. The school district had alleged that addictive design features on Instagram and Facebook harmed students and had sought US$60 million in damages. Meta denied wrongdoing as part of the settlement.<a href="https://embed.wionews.com/videos/meta-hit-with-1-4-trillion-penalty-claim-by-us-states-1783449721227/embed">https://embed.wionews.com/videos/meta-hit-with-1-4-trillion-penalty-claim-by-us-states-1783449721227/embed</a><h2>Social media addiction: The case of Kaley Glenn-Mills</h2>In March 2026, a Los Angeles jury in the K.G.M. v. Meta Platforms, Inc. et al case found Meta and Google liable for negligently designing addictive features on Instagram and YouTube that harmed the plaintiff, Kaley Glenn-Mills. The verdict is viewed as one of the first successful individual social media addiction cases to go before a jury.The lawsuit centred on platform features including infinite scroll, algorithmic recommendations, autoplay and other engagement-driven design elements. The plaintiff's lawyers argued that these features contributed to compulsive use, leading to anxiety, depression, body dysmorphia and other mental health problems.Also read: <a href="https://www.wionews.com/technology/instagram-will-now-let-anyone-generate-deepfake-of-your-face-unless-you-opt-out-right-now-1783477170729">Instagram will now let anyone generate deepfake of your face, unless you opt out right now</a>&nbsp;According to court testimony, Glenn-Mills began using YouTube at the age of six and Instagram at the age of nine. Her lawyers argued that years of compulsive social media use contributed to severe mental health difficulties, including depression, self-harm, suicidal ideation, eating disorders and repeated psychiatric hospitalisations during her adolescence.&nbsp;The jury awarded US$6 million in total damages — US$3 million in compensatory damages and US$3 million in punitive damages. It apportioned 70 per cent of the liability , or US$4.2 million, to Meta and 30 per cent, or US$1.8 million, to Google.Meta said it "respectfully disagrees" with the verdict. The company argued that teenage mental health is "profoundly complex" and cannot be attributed to a single app. It added: "We will continue to defend ourselves vigorously as every case is different, and we remain confident in our record of protecting teens online."Also in March, a New Mexico jury ordered Meta to pay US$375 million in civil penalties after finding that it violated the state's consumer protection law by misleading consumers about the safety of its platforms and concealing what it knew about child sexual exploitation risks on Facebook and Instagram. &nbsp;The case was brought by New Mexico Attorney General Raúl Torrez. The jury awarded the maximum statutory penalty of US$5,000 per violation, amounting to US$375 million.Both the California verdict in K.G.M. v. Meta Platforms, Inc. et al. and the New Mexico judgment have been appealed by Meta.<h2>Social media, Big Tech and consumers: Several cases going on</h2>Meta, Google and other Big Tech companies are facing thousands of lawsuits consolidated in federal multidistrict litigation, as well as separate actions brought by more than 40 US state attorneys general, school districts and individual plaintiffs.The main allegation is that social media platforms were designed to foster compulsive use among children and teenagers and failed to adequately warn users about the associated risks.

Meta faces trial in Tennessee over claims Instagram was designed to addict teenagers. This case adds to growing US lawsuits over addictive algorithms, teen mental health and platform safety. Here is why it matters

A US court has begun a trial against Meta Platforms, the parent company of Facebook, Instagram and WhatsApp, over allegations that it deliberately designed addictive features targeting teenagers. Jury selection began in Nashville, Tennessee, on Monday (Jul 20), in a lawsuit brought by the Tennessee Attorney General under the state's consumer protection law. 'Instagram, the addiction machine' Tennessee Attorney General Jonathan Skrmetti accused Meta of knowingly designing Instagram as an "addiction machine" that drives compulsive use among teenagers. The lawsuit alleged that Meta failed to disclose its own internal research on the platform's harms to young users and misled the public about its safety. According to the complaint, Meta continued deploying features it allegedly knew were harmful, while downplaying the associated risks. Meta chief executive Mark Zuckerberg previously testified that the company's algorithms are not intentionally designed to make young users addicted and that Meta has invested in safety features for teenagers. The trial is expected to last six to seven weeks. If Meta is found liable, the court could impose civil penalties of up to US$1,000 for each violation of Tennessee's Consumer Protection Act. it could also order changes to Instagram's design and practices. Also read: 'Broken AI, useless products': Mark Zuckerberg might be steering Meta towards a disaster Meta is facing several cases across the US The Tennessee case overlaps with a growing number of lawsuits against Meta alleging that its platforms were deliberately designed to maximise engagement at the expense of young users' mental health. In May this year, Meta settled a lawsuit brought by Breathitt County Schools in Kentucky for US$9 million. The school district had alleged that addictive design features on Instagram and Facebook harmed students and had sought US$60 million in damages. Meta denied wrongdoing as part of the settlement. https://embed.wionews.com/videos/meta-hit-with-1-4-trillion-penalty-claim-by-us-states-1783449721227/embed Social media addiction: The case of Kaley Glenn-Mills In March 2026, a Los Angeles jury in the K.G.M. v. Meta Platforms, Inc. et al case found Meta and Google liable for negligently designing addictive features on Instagram and YouTube that harmed the plaintiff, Kaley Glenn-Mills. The verdict is viewed as one of the first successful individual social media addiction cases to go before a jury. The lawsuit centred on platform features including infinite scroll, algorithmic recommendations, autoplay and other engagement-driven design elements. The plaintiff's lawyers argued that these features contributed to compulsive use, leading to anxiety, depression, body dysmorphia and other mental health problems. Also read: Instagram will now let anyone generate deepfake of your face, unless you opt out right now According to court testimony, Glenn-Mills began using YouTube at the age of six and Instagram at the age of nine. Her lawyers argued that years of compulsive social media use contributed to severe mental health difficulties, including depression, self-harm, suicidal ideation, eating disorders and repeated psychiatric hospitalisations during her adolescence. The jury awarded US$6 million in total damages — US$3 million in compensatory damages and US$3 million in punitive damages. It apportioned 70 per cent of the liability , or US$4.2 million, to Meta and 30 per cent, or US$1.8 million, to Google. Meta said it "respectfully disagrees" with the verdict. The company argued that teenage mental health is "profoundly complex" and cannot be attributed to a single app. It added: "We will continue to defend ourselves vigorously as every case is different, and we remain confident in our record of protecting teens online." Also in March, a New Mexico jury ordered Meta to pay US$375 million in civil penalties after finding that it violated the state's consumer protection law by misleading consumers about the safety of its platforms and concealing what it knew about child sexual exploitation risks on Facebook and Instagram. The case was brought by New Mexico Attorney General Raúl Torrez. The jury awarded the maximum statutory penalty of US$5,000 per violation, amounting to US$375 million. Both the California verdict in K.G.M. v. Meta Platforms, Inc. et al. and the New Mexico judgment have been appealed by Meta. Social media, Big Tech and consumers: Several cases going on Meta, Google and other Big Tech companies are facing thousands of lawsuits consolidated in federal multidistrict litigation, as well as separate actions brought by more than 40 US state attorneys general, school districts and individual plaintiffs. The main allegation is that social media platforms were designed to foster compulsive use among children and teenagers and failed to adequately warn users about the associated risks.

Meta faces trial in Tennessee over claims Instagram was designed to addict teenagers. This case adds to growing US lawsuits over addictive algorithms, teen mental health and platform safety. Here is why it matters&nbsp;

Addictive algorithms: Meta trial opens in Tennessee as Instagram accused of 'addiction machine’ design

technology

Technology, News on Technology, stories on Technology, articles on Technology, Technology stories at Wion, Technology news, Technology latest news

technology

'Scary' Rogue AI on the loose? OpenAI agent hit another tech firm's customer, suggests report

Trump administration bans foreign-made humanoid robots. Here's why

Why did Meta remove Modi's Facebook video on paper leaks? Tech company says...

Why are brands turning to AI influencers for their marketing campaigns?

What is AI singularity? Sam Altman says the world has already entered it

Elon Musk’s tunnel company reportedly wants $4 billion from investors

OpenAI wants ChatGPT everywhere, including the smart glasses you wear

Can AI deliver human-like customer experience at scale? Experts weigh in

What is Prentis? AI lab reportedly seeking $100 million after launching in April

‘I’m not racist’: Elon Musk points to half-Indian partner Shivon Zilis

Claude Opus 5 is here: Fable-like intelligence at half the cost

Google likely to stop these Android phones from downloading apps & games

'Nvidia is digging its own grave': DeepSeek CEO backs China's AI chip push

'Not the opposite': Jensen Huang says Taiwan saved US chip industry

AI could surpass combined human intelligence by 2031, Elon Musk predicts

Anthropic wants 2 GW of AMD AI chips. Here's why the deal matters for Nvidia

Microsoft has a new plan for AI: Use expensive frontier models only when needed

Urgent UN warnings: time to tame killer robots is running out

Google built a cyber-warfare AI that only governments can use

Moonshot AI faces US sanctions over ‘distillation attacks’ to exploit Anthropic

Nvidia says its next AI superchip is 10 times more efficient than Blackwell

Anthropic's $1.5 billion settlement explained: What authors get and what changes

TSMC's reported chip price hike could raise smartphone prices. Here's why

OpenAI promotes Indian-origin engineer Uday Ruddarraju to CTO, Compute

OpenAI’s AI agent gone rogue? Company reveals chilling hack at startup

Meta trial opens in Tennessee as Instagram 'addiction machine' slammed