Anthropic has filed a lawsuit against the United States Government after being barred from certain military-related engagements, intensifying tensions over the role of artificial intelligence in defense operations. The dispute highlights growing friction between tech companies and government agencies over AI safety, national security restrictions, and the use of advanced models in military systems.


Anthropic Sues US Government Over Military Ban; Dispute Escalates

U.S. President Donald Trump has urged the government to remove Anthropic from its AI projects, amid ongoing controversy over the company’s inclusion on a Pentagon AI blacklist. In response, Anthropic filed a lawsuit against the Department of Defense, challenging its exclusion from AI initiatives.


Trump Urges Govt to Drop Anthropic's AI | Anthropic Sues Pentagon Over AI Blacklist

AI firm Anthropic is preparing a legal challenge after the Pentagon reportedly labeled the company a potential supply-chain security risk. The dispute could trigger a major legal battle over AI technology, defence procurement, and national security oversight in the United States. watch this report for more details!


Tech Firm Anthropic Sues Donald Trump Administration Over Pentagon Blacklist

Let me give you both the most pessimistic and optimistic views of where the middle classes of countries like the US and India will end up if AI adoption does not go beyond merely asking ChatGPT for answers, or Grok for funny images and videos. Companies at the forefront of artificial intelligence are warning about how jobs across industries could be wiped out by AI. This could potentially lead to a middle-class collapse in these countries if action does not match the situation. How bad can it get? And what is the way out?<h2>A disturbing discussion led by AI leaders on the impending fall of jobs</h2>The starting point for this article is a disturbing series of discussions on a future fall in jobs, ironically led by AI leaders themselves. Anthropic CEO Dario Amodei is ‘leading the charge’ in bluntly saying how traditional jobs, including ‘secure’ jobs in the legacy IT sector, are going to be wiped out. In recent interviews and public remarks, Amodei indicated that up to 50 per cent of entry-level work in fields such as law, consulting, and finance could be disrupted or replaced by AI systems within the next one to five years.Reacting to this, Damian Player, CEO of the AI consulting firm Agent Integrator, recently posted on X: “This is f***ing insane! The CEO of Anthropic said what most C-suites only say behind closed doors… this is the head of one of the biggest AI companies alive telling ordinary people to wake up.”<h2>A dystopian scenario if India does not adapt to AI beyond current levels</h2>The situation being described below is only a simulation of the future, taking cues from the ongoing K-shaped growth in the US, underway since the Covid pandemic. K-shaped growth is when certain sections of society and the economy show exponential growth, while other sections race to the bottom, making the resultant image resemble the letter K. In a future scenario where AI eats up jobs, large sections of Indians could be pushed towards poverty. This could happen as early as the next five years in the absence of even foundational-level skills in AI.<a href="https://www.wionews.com/videos/anthropic-vows-court-battle-after-pentagon-flags-ai-firm-as-security-risk-wion-1772880262306/embed">https://www.wionews.com/videos/anthropic-vows-court-battle-after-pentagon-flags-ai-firm-as-security-risk-wion-1772880262306/embed</a><h2>What is the level of AI adoption in India? Identifying the problem</h2>It is quite possible that a large majority of Indians have never actively used AI tools in a meaningful way. Even among the digitally savvy population, many have little knowledge of AI apart from using the free version of ChatGPT and the like. Contrast this with the current world, where agentic and ‘copilot’ systems such as Anthropic’s Claude integrations and automated workflow tools are slowly entering enterprise environments.Only a small fraction of users subscribe to paid AI platforms that typically cost around ₹2,000 per month or more. An even smaller proportion use coding interfaces, APIs, or automated workflows to interact with AI systems.<figure class="image"><img src="https://cdn1.wionews.com/prod/wion/images/2026/20260309/image-1773050725928.jpg"></figure>An infographic showing the level of AI adoption in the world (Image Courtesy: LinkedIn/ Benjamin T, co-Founder at Aiso)<h2>Why K-shaped growth in the US should be a wake-up call to India</h2>The K-shaped growth pattern is already visible in the US economy. In the current scenario, this is partly a function of AI adoption too. Many Americans are forcing themselves to learn AI use cases in order to remain competitive in the workplace. In terms of AI adoption, analysts often describe four levels of participation, known as AI maturity models: Elite, Augmented, Middle, and Struggling. It is projected that around 30 per cent of American citizens could remain in the Elite and Augmented categories, while 50 per cent could fall into the Middle group, and about 20 per cent may Struggle to keep up.Also read: <a href="https://www.wionews.com/photos/-end-of-ias-7-reasons-why-ai-will-make-upsc-exam-useless-in-2-years-1772938793987">'End of IAS': 7 reasons why AI will make UPSC exam useless in 2 years</a><h2>Indians are ‘consumers’ of AI, and that is prelude to potential digital colonisation</h2>There is a clear danger in Indians becoming consumers of international AI companies, essentially paying for subscription packages ranging anywhere from ₹2,000 or upwards to use their AI models. This is because India has yet to build globally competitive AI tools at scale, in spite of several initiatives at the government (India AI mission) and corporate (Infosys, TCS and others incoporating AI in workflows) levels. Even ‘localisation efforts’ are largely being driven by global technology companies such as Google, Microsoft, and OpenAI.&nbsp;In several ways, this echoes patterns seen during colonial economic structures, when the British Raj turned large sections of the Indian population into consumers of value-added goods produced elsewhere.If this trajectory continues, India could risk a form of neo-digital colonisation, where countries ahead of the AI curve — particularly the US and China — dominate the virtual and technological landscape.Indians would, in effect, end up paying US or Chinese companies for AI services, revenue that could ultimately support these countries' domestic technology ecosystems and economic programmes.Also read: <a href="https://www.wionews.com/business-economy/at-92-per-cent-india-s-ai-adoption-rate-highest-in-the-asia-pacific-region-key-figures-1762333628812">India AI Impact Summit: $300 billion worth of AI investments committed in India, say reports</a>&nbsp;<figure class="image"><img src="https://cdn1.wionews.com/prod/wion/images/2026/20260309/image-1773051163876.jpeg"></figure>A three-year price drop of Nifty IT stocks&nbsp;<h2>Indian IT stocks crash, HNIs trickle out: A fever indicating worse possible outcomes in the near future</h2>Not too far away in the future, full-stack AI companies could enter India aggressively, resulting in a dramatic rise in layoffs and a stock market correction led by the decline of traditional IT service firms. And in some ways, the warning signs are already visible when it comes to IT stocks.The Nifty IT index has already shown significant volatility when viewed across recent multi-year patterns. At one point, it fell around 30–34 per cent from its peak within a span of several months (see image above).For some observers, this evokes memories of the early stages of the 2007 downturn that eventually led to the global financial crisis triggered by the US mortgage market collapse.A prolonged slowdown could potentially spill over into banking and other industries.&nbsp;Layoffs, which have already been seen periodically in the global technology sector, could hit traditional IT jobs in India particularly hard.A major downturn could also affect the Indian rupee, which has gradually weakened against the US dollar over the past decade.Another indicator often cited in this context is the growing number of high-net-worth individuals (HNIs) exploring migration. According to various global wealth reports, several thousand millionaires are estimated to leave India annually, often relocating to countries such as the UK, United Arab Emirates, Singapore, Australia, and Canada.<h2>So what is the way out? Adopt AI or die</h2>Damian Player’s post that I mentioned earlier, also offers a possible cure for the problem. To paraphrase him, the good news is that the playbook has not changed.There is still a pathway for AI adoption for those who are currently ‘AI-illiterate’. As Player said, pick an industry, learn the workflows. Become the ‘AI person’ these companies need.As Player puts it, you do not need a degree. “You sure as hell don’t need to be a McKinsey consultant. You need Claude, a vertical, and the hunger to learn a business better than anyone else. God speed! I’ll be rooting for ya,” Player wrote.Godspeed is all I can say to those standing at the edge of the AI cliff — and staring downwards.

AI company leaders are warning that generative AI could disrupt entry-level white-collar jobs, deepen inequality and cause economic collapse. If adoption lags, countries like India risk a widening AI divide shaping the future of work and the middle class. Here are the worst and best cases

Let me give you both the most pessimistic and optimistic views of where the middle classes of countries like the US and India will end up if AI adoption does not go beyond merely asking ChatGPT for answers, or Grok for funny images and videos. Companies at the forefront of artificial intelligence are warning about how jobs across industries could be wiped out by AI. This could potentially lead to a middle-class collapse in these countries if action does not match the situation. How bad can it get? And what is the way out? A disturbing discussion led by AI leaders on the impending fall of jobs The starting point for this article is a disturbing series of discussions on a future fall in jobs, ironically led by AI leaders themselves. Anthropic CEO Dario Amodei is ‘leading the charge’ in bluntly saying how traditional jobs, including ‘secure’ jobs in the legacy IT sector, are going to be wiped out. In recent interviews and public remarks, Amodei indicated that up to 50 per cent of entry-level work in fields such as law, consulting, and finance could be disrupted or replaced by AI systems within the next one to five years. Reacting to this, Damian Player, CEO of the AI consulting firm Agent Integrator, recently posted on X: “This is f***ing insane! The CEO of Anthropic said what most C-suites only say behind closed doors… this is the head of one of the biggest AI companies alive telling ordinary people to wake up.” A dystopian scenario if India does not adapt to AI beyond current levels The situation being described below is only a simulation of the future, taking cues from the ongoing K-shaped growth in the US, underway since the Covid pandemic. K-shaped growth is when certain sections of society and the economy show exponential growth, while other sections race to the bottom, making the resultant image resemble the letter K. In a future scenario where AI eats up jobs, large sections of Indians could be pushed towards poverty. This could happen as early as the next five years in the absence of even foundational-level skills in AI. https://www.wionews.com/videos/anthropic-vows-court-battle-after-pentagon-flags-ai-firm-as-security-risk-wion-1772880262306/embed What is the level of AI adoption in India? Identifying the problem It is quite possible that a large majority of Indians have never actively used AI tools in a meaningful way. Even among the digitally savvy population, many have little knowledge of AI apart from using the free version of ChatGPT and the like. Contrast this with the current world, where agentic and ‘copilot’ systems such as Anthropic’s Claude integrations and automated workflow tools are slowly entering enterprise environments. Only a small fraction of users subscribe to paid AI platforms that typically cost around ₹2,000 per month or more. An even smaller proportion use coding interfaces, APIs, or automated workflows to interact with AI systems. An infographic showing the level of AI adoption in the world (Image Courtesy: LinkedIn/ Benjamin T, co-Founder at Aiso) Why K-shaped growth in the US should be a wake-up call to India The K-shaped growth pattern is already visible in the US economy. In the current scenario, this is partly a function of AI adoption too. Many Americans are forcing themselves to learn AI use cases in order to remain competitive in the workplace. In terms of AI adoption, analysts often describe four levels of participation, known as AI maturity models: Elite, Augmented, Middle, and Struggling. It is projected that around 30 per cent of American citizens could remain in the Elite and Augmented categories, while 50 per cent could fall into the Middle group, and about 20 per cent may Struggle to keep up. Also read: 'End of IAS': 7 reasons why AI will make UPSC exam useless in 2 years Indians are ‘consumers’ of AI, and that is prelude to potential digital colonisation There is a clear danger in Indians becoming consumers of international AI companies, essentially paying for subscription packages ranging anywhere from ₹2,000 or upwards to use their AI models. This is because India has yet to build globally competitive AI tools at scale, in spite of several initiatives at the government (India AI mission) and corporate (Infosys, TCS and others incoporating AI in workflows) levels. Even ‘localisation efforts’ are largely being driven by global technology companies such as Google, Microsoft, and OpenAI. In several ways, this echoes patterns seen during colonial economic structures, when the British Raj turned large sections of the Indian population into consumers of value-added goods produced elsewhere. If this trajectory continues, India could risk a form of neo-digital colonisation, where countries ahead of the AI curve — particularly the US and China — dominate the virtual and technological landscape. Indians would, in effect, end up paying US or Chinese companies for AI services, revenue that could ultimately support these countries' domestic technology ecosystems and economic programmes. Also read: India AI Impact Summit: $300 billion worth of AI investments committed in India, say reports A three-year price drop of Nifty IT stocks Indian IT stocks crash, HNIs trickle out: A fever indicating worse possible outcomes in the near future Not too far away in the future, full-stack AI companies could enter India aggressively, resulting in a dramatic rise in layoffs and a stock market correction led by the decline of traditional IT service firms. And in some ways, the warning signs are already visible when it comes to IT stocks. The Nifty IT index has already shown significant volatility when viewed across recent multi-year patterns. At one point, it fell around 30–34 per cent from its peak within a span of several months (see image above). For some observers, this evokes memories of the early stages of the 2007 downturn that eventually led to the global financial crisis triggered by the US mortgage market collapse. A prolonged slowdown could potentially spill over into banking and other industries. Layoffs, which have already been seen periodically in the global technology sector, could hit traditional IT jobs in India particularly hard. A major downturn could also affect the Indian rupee, which has gradually weakened against the US dollar over the past decade. Another indicator often cited in this context is the growing number of high-net-worth individuals (HNIs) exploring migration. According to various global wealth reports, several thousand millionaires are estimated to leave India annually, often relocating to countries such as the UK, United Arab Emirates, Singapore, Australia, and Canada. So what is the way out? Adopt AI or die Damian Player’s post that I mentioned earlier, also offers a possible cure for the problem. To paraphrase him, the good news is that the playbook has not changed. There is still a pathway for AI adoption for those who are currently ‘AI-illiterate’. As Player said, pick an industry, learn the workflows. Become the ‘AI person’ these companies need. As Player puts it, you do not need a degree. “You sure as hell don’t need to be a McKinsey consultant. You need Claude, a vertical, and the hunger to learn a business better than anyone else. God speed! I’ll be rooting for ya,” Player wrote. Godspeed is all I can say to those standing at the edge of the AI cliff — and staring downwards.

AI company leaders are warning that generative AI could disrupt entry-level white-collar jobs, deepen inequality and cause economic collapse. If adoption lags, countries like India risk a widening AI divide shaping the future of work and the middle class. Here are the worst and best cases

This is f***ing insane!: AI’s dystopian future is K-shaped, and could decimate the middle classes from India to the US

Anthropic Vows Court Battle After Pentagon Flags AI Firm As Security Risk | WION 

Anthropic plans to legally challenge the Pentagon’s “supply chain risk” designation, which could affect government contracts. The AI firm argues the label is unwarranted and threatens its operations. This move highlights growing tensions between AI developers and U.S. defense authorities over regulation and compliance.

Anthropic  Vs Pentagon: Anthropic to Challenge Pentagon “Supply Chain Risk” Label

Several U.S. government departments, including the State Department, Treasury and Health and Human Services, have reportedly shifted away from AI systems developed by Anthropic following a dispute over access, safeguards, and potential military applications.

Anthropic has publicly rejected involvement in mass surveillance programs and the development of autonomous weapons, saying its policies restrict such use cases.


Trump Announces Six-Month AI Phase-Out: Legal Doubts Cloud Supply Chain Ban

AI company Anthropic has reportedly declined a Pentagon ultimatum seeking unconditional use of its artificial intelligence systems by the US military. The development signals tensions between major AI firms and defence authorities over how advanced AI technologies should be deployed, particularly in military and national security contexts.


AI: Anthropic Defies Pentagon, Refuses Unconditional US Military AI Access | WION 

The Pentagon has issued a strict ultimatum to AI company Anthropic over the use of its artificial intelligence models in defense applications. U.S. Defense Secretary Pete Hegseth warned that failure to comply with Pentagon guidelines could lead to contract termination. The move underscores growing scrutiny over AI in military operations and the need for strict oversight of advanced technologies with national security implications.


Pentagon Has Issued a Sharp Ultimatum to AI Firm Anthropic on Military Use of Its Models

AI safety firm Anthropic has publicly accused three Chinese AI labs—DeepSeek, Moonshot AI, and MiniMax—of orchestrating "industrial-scale" campaigns to illicitly copy the capabilities of its Claude AI model. The operation involved using approximately 24,000 fraudulent accounts to conduct over 16 million exchanges with Claude, violating Anthropic’s terms of service and bypassing regional access restrictions. 


Claude Misuse Allegations: Fake Accounts Used To Tap AI Capabilities

A brief onstage moment between OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei drew attention during the India AI Impact Summit 2026 in New Delhi.&nbsp;During a group photograph with Prime Minister Narendra Modi and other global technology leaders, several attendees were seen holding hands and raising them for the cameras. However, social media users quickly noticed that Altman and Amodei did not hold hands. Instead, both raised their fists separately, maintaining distance between them.&nbsp;The moment soon went viral, with many online speculating about tensions between the two AI leaders.&nbsp;Prime Minister Narendra Modi stood at the centre of the photograph. Google CEO Sundar Pichai stood on his right, while Altman stood on his left. Amodei was positioned next to Altman.&nbsp;<h2>Sam Altman Responds: ‘I Was Confused’</h2>Hours after the image spread online, Sam Altman said about the incident during a closed-door interaction with few reporters.&nbsp;“I was sort of confused and didn’t know what I was supposed to do,” Altman said. He explained that Prime Minister Modi had taken his hand and raised it, but he was unsure about the intended gesture for the photograph.&nbsp;Altman added, “Modi grabbed my hand and put it up, and I wasn’t sure what we were supposed to be doing.”&nbsp;As of now there was no comment from Anthropic CEO Dario Amodei on the incident.&nbsp;<h2>OpenAI and Anthropic Rivalry</h2>Sam Altman, 40, and Dario Amodei, 43, previously worked together at OpenAI until 2019. According to earlier reports, Amodei left OpenAI in 2020 along with several senior researchers following disagreements over the company’s direction. He later co-founded Anthropic with his sister Daniela Amodei.&nbsp;In 2023, when OpenAI’s board temporarily removed Altman as CEO, reports suggested that Amodei was approached as a possible replacement. He declined the offer. Altman was reinstated as CEO within days.&nbsp;Today, OpenAI and Anthropic are among the leading artificial intelligence firms globally. According to Bloomberg estimates cited in reports, OpenAI is valued at around $500 billion, while Anthropic is valued at approximately $380 billion.&nbsp;The rivalry has also played out publicly. Anthropic recently aired a high-profile Super Bowl advertisement that criticised the idea of intrusive AI advertising. The ad suggested that some AI companies might insert advertisements directly into chatbot responses.&nbsp;In response, Altman posted on X that he “laughed” at the ad but described it as “clearly dishonest.” He said OpenAI’s principle was not to insert ads directly into chatbot responses. He also claimed that “more Texans use ChatGPT for free than total people use Claude in the US.”&nbsp;<h2>India AI Impact Summit them brought together</h2>The India AI Impact Summit 2026 brought together global AI leaders, policymakers and industry executives to discuss artificial intelligence governance, infrastructure and innovation. With OpenAI and Anthropic both expanding globally, including in India, public appearances of their CEOs attract close attention.&nbsp;While the viral moment sparked speculation online, Altman’s clarification suggests there was no deliberate snub, but rather uncertainty during a coordinated stage gesture.&nbsp;&nbsp;

A viral moment at the India AI Impact Summit showed Sam Altman and Dario Amodei not holding hands during a group photo with PM Modi. Altman later said he was “confused” about the gesture. The clip sparked speculation amid OpenAI-Anthropic rivalry.


A brief onstage moment between OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei drew attention during the India AI Impact Summit 2026 in New Delhi. During a group photograph with Prime Minister Narendra Modi and other global technology leaders, several attendees were seen holding hands and raising them for the cameras. However, social media users quickly noticed that Altman and Amodei did not hold hands. Instead, both raised their fists separately, maintaining distance between them. The moment soon went viral, with many online speculating about tensions between the two AI leaders. Prime Minister Narendra Modi stood at the centre of the photograph. Google CEO Sundar Pichai stood on his right, while Altman stood on his left. Amodei was positioned next to Altman. Sam Altman Responds: ‘I Was Confused’ Hours after the image spread online, Sam Altman said about the incident during a closed-door interaction with few reporters. “I was sort of confused and didn’t know what I was supposed to do,” Altman said. He explained that Prime Minister Modi had taken his hand and raised it, but he was unsure about the intended gesture for the photograph. Altman added, “Modi grabbed my hand and put it up, and I wasn’t sure what we were supposed to be doing.” As of now there was no comment from Anthropic CEO Dario Amodei on the incident. OpenAI and Anthropic Rivalry Sam Altman, 40, and Dario Amodei, 43, previously worked together at OpenAI until 2019. According to earlier reports, Amodei left OpenAI in 2020 along with several senior researchers following disagreements over the company’s direction. He later co-founded Anthropic with his sister Daniela Amodei. In 2023, when OpenAI’s board temporarily removed Altman as CEO, reports suggested that Amodei was approached as a possible replacement. He declined the offer. Altman was reinstated as CEO within days. Today, OpenAI and Anthropic are among the leading artificial intelligence firms globally. According to Bloomberg estimates cited in reports, OpenAI is valued at around $500 billion, while Anthropic is valued at approximately $380 billion. The rivalry has also played out publicly. Anthropic recently aired a high-profile Super Bowl advertisement that criticised the idea of intrusive AI advertising. The ad suggested that some AI companies might insert advertisements directly into chatbot responses. In response, Altman posted on X that he “laughed” at the ad but described it as “clearly dishonest.” He said OpenAI’s principle was not to insert ads directly into chatbot responses. He also claimed that “more Texans use ChatGPT for free than total people use Claude in the US.” India AI Impact Summit them brought together The India AI Impact Summit 2026 brought together global AI leaders, policymakers and industry executives to discuss artificial intelligence governance, infrastructure and innovation. With OpenAI and Anthropic both expanding globally, including in India, public appearances of their CEOs attract close attention. While the viral moment sparked speculation online, Altman’s clarification suggests there was no deliberate snub, but rather uncertainty during a coordinated stage gesture.

A viral moment at the India AI Impact Summit showed Sam Altman and Dario Amodei not holding hands during a group photo with PM Modi. Altman later said he was “confused” about the gesture. The clip sparked speculation amid OpenAI-Anthropic rivalry.

'Didn't know what was happening': Here’s what Sam Altman said on his awkward exchange with Anthropic CEO Dario Amodei

Anthropic's artificial-intelligence model Claude was used in the U.S. military's operation to capture former Venezuelan President Nicolas Maduro, the Wall Street Journal reported on Friday, citing people familiar with the matter.
Claude's deployment came via Anthropic's partnership with data firm Palantir Technologies (PLTR.O), opens new tab, whose platforms are widely used by the Defense Department and federal law enforcement, the report added.

WSJ: Anthropic's Claude AI Deployed in Maduro Raid

As we track the race for ai dominance, Anthropic has quietly emerged as a leader in the AI boom. Experts say its advantage comes from focus, not flash.


How Anthropic Surged Ahead in the Race for AI Dominance

The rivalry between OpenAI and Anthropic sharpened this week after both companies released new artificial intelligence models on the same day, highlighting growing competition in advanced AI development.Anthropic launched Claude Opus 4.6, while OpenAI unveiled GPT-5.3-Codex, with both companies positioning their models as stronger tools for coding and workplace productivity.<h2>What Anthropic announced with Claude Opus 4.6</h2>Anthropic said Claude Opus 4.6 is designed to handle longer and more complex tasks in a single session. As per the company, the model offers an expanded context window, allowing it to process large documents and detailed projects more efficiently.Anthropic claims the update improves performance in office productivity and software development tasks. The release follows feedback from users who compared earlier Anthropic and OpenAI models, often pointing to differences in speed and problem-solving depth.Speaking on the TBPN podcast, Anthropic researcher Sholto Douglas said the company focused on making its model stronger at handling difficult problems, even if that required more careful reasoning.

<blockquote class="twitter-tweet">Introducing Claude Opus 4.6. Our smartest model got an upgrade. Opus 4.6 plans more carefully, sustains agentic tasks for longer, operates reliably in massive codebases, and catches its own mistakes. It’s also our first Opus-class model with 1M token context in beta. <a href="https://t.co/L1iQyRgT9x">pic.twitter.com/L1iQyRgT9x</a>&mdash; Claude (@claudeai) <a href="https://twitter.com/claudeai/status/2019467372609040752?ref_src=twsrc%5Etfw">February 5, 2026</a></blockquote> <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>

<h2>OpenAI’s announced GPT-5.3-Codex</h2>On the same day, OpenAI released GPT-5.3-Codex, a model built specifically for coding and software management. The company said the model runs faster, uses fewer computing resources, and can manage complex programming tasks using plain English instructions.OpenAI also introduced a standalone Codex desktop app alongside the model, aiming to make AI-assisted coding easier for developers.OpenAI CEO Sam Altman said users are increasingly moving towards workflows where they manage teams of AI agents rather than writing every instruction themselves.&nbsp;

<blockquote class="twitter-tweet">GPT-5.3-Codex is now available in Codex. You can just build things.<a href="https://t.co/dyBiIQXGx1">https://t.co/dyBiIQXGx1</a>&mdash; OpenAI (@OpenAI) <a href="https://twitter.com/OpenAI/status/2019474152743223477?ref_src=twsrc%5Etfw">February 5, 2026</a></blockquote> <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>

&nbsp;Rivalry&nbsp;The competition between OpenAI and Anthropic dates back to 2021, when former OpenAI researchers left to form Anthropic, focusing on building more controlled and safety-focused AI systems.This week also proved significant for Anthropic in other ways. Its release of industry-specific AI tools raised concerns on Wall Street about the impact of AI on traditional software companies, triggering a market sell-off.Anthropic also released advertisements promoting its chatbot Claude as ad-free, a clear contrast to OpenAI’s recent decision to introduce ads for free ChatGPT users. OpenAI announced earlier this year that advertising would help fund its growing operational costs.Altman pushed back against criticism, saying ads would be clearly labelled and would not influence chatbot responses. He added that inserting ads into AI answers would damage user trust.&nbsp;What this means for the AI industryThe same-day launches underline how fast competition is moving in the AI sector, especially in tools aimed at developers and professionals. As companies race to improve speed, accuracy and efficiency, users are increasingly comparing models side by side.With OpenAI and Anthropic continuing to release frequent updates, experts say the AI market is shifting from novelty to real-world productivity, where performance, cost and trust are becoming key deciding factors.

OpenAI and Anthropic escalated their AI rivalry by launching new models on the same day. Anthropic unveiled Claude Opus 4.6, focused on long, complex tasks, while OpenAI released GPT-5.3-Codex for faster, efficient AI-powered coding and productivity.

The rivalry between OpenAI and Anthropic sharpened this week after both companies released new artificial intelligence models on the same day, highlighting growing competition in advanced AI development. Anthropic launched Claude Opus 4.6, while OpenAI unveiled GPT-5.3-Codex, with both companies positioning their models as stronger tools for coding and workplace productivity. What Anthropic announced with Claude Opus 4.6 Anthropic said Claude Opus 4.6 is designed to handle longer and more complex tasks in a single session. As per the company, the model offers an expanded context window, allowing it to process large documents and detailed projects more efficiently. Anthropic claims the update improves performance in office productivity and software development tasks. The release follows feedback from users who compared earlier Anthropic and OpenAI models, often pointing to differences in speed and problem-solving depth. Speaking on the TBPN podcast, Anthropic researcher Sholto Douglas said the company focused on making its model stronger at handling difficult problems, even if that required more careful reasoning.

OpenAI and Anthropic escalated their AI rivalry by launching new models on the same day. Anthropic unveiled Claude Opus 4.6, focused on long, complex tasks, while OpenAI released GPT-5.3-Codex for faster, efficient AI-powered coding and productivity.

OpenAI and Anthropic release new AI models the same day as their rivalry intensifies

Markets had its worst software selloff in years on Monday, erasing $285 billion in market value, after Anthropic released plugins that let its Claude AI handle legal contracts, sales research, and financial analysis without human supervision.The storm hit everything from legacy software giants to Indian outsourcers. Thomson Reuters lost $8.2 billion in a single day, its steepest drop ever. Infosys fell 8%, TCS 6%, and the Nifty IT index shed RS1.9 lakh crore, dropping below RS30 lakh crore for the first time in months.What tensed investors wasn’t the technology itself. It was realizing that companies might soon pay for one AI agent instead of ten software licenses, or a hundred consulting hours.What changed on January 30 Anthropic launched Claude Cowork on January 12 as a desktop tool for non-coders. Users could tell it to draft documents, analyze spreadsheets, or organize files. Handy, but not revolutionary.Then came the plugins.On January 30, the company open-sourced 11 specialized add-ons on GitHub: contract reviews, NDA screening, compliance checks, legal briefings, sales prospecting, CRM integration, financial modeling, marketing automation.The AI could now execute multi-step workflows autonomously, plan tasks, spawn sub-processes, and deliver results. “When we released Claude Code, we expected developers to use it for coding,” Anthropic explained in a blog post. “They did, and then quickly began using it for almost everything else. This prompted us to build Cowork: a simpler way for anyone, not just developers, to work with Claude in the very same way.”That “same way” is the problem for software companies. Claude doesn’t just assist anymore. It executes.<h2>The billable-hour apocalypse</h2>Traders coined a new term Monday: “SaaSpocalypse.” Goldman Sachs’ software basket plunged 6%, matching its worst day since April’s tariff selloff. The Nasdaq 100 dropped 2.4% before recovering to close 1.6% down. Financial services stocks tumbled 7%.Indian IT services took the hardest hit. These companies built empires on delivering skilled labor for tasks AI can now automate. HCL Technologies and LTIMindtree fell 5-7%, Wipro dropped 4%. The Nifty IT index posted its worst month since October 2008, down 15% in January alone.<h2>Panic or prophecy?</h2>Industry experts watching the carnage split on whether markets overreacted or saw the future clearly.Adv. Varun Singh, founder and managing partner at Foresight Law Offices India, believes the selloff reflects shifting value perception rather than immediate extinction. “The way the market’s reacted demonstrates the speed with which AI programs are altering people’s ideas of what software should cost,” Varun said.“The substantial decline suggests concern that AI can consolidate various SaaS roles into one smart unit, leading to fewer tools being necessary to do specialized tasks.”But he sees evolution, not apocalypse.“This doesn’t mean big, established SaaS firms will suddenly become obsolete. The core systems remain deeply implemented and critical to business success across verticals. The actual disruption will materialize in value distribution, with businesses favoring platforms that combine legal, operational, and strategic intelligence. Consequently, incumbents will be forced to either evolve or risk margin compression rather than outright extinction.”Suyash Lahoti, partner at Wit &amp; Chai group, believes investors missed the point. The threat isn’t software becoming irrelevant overnight, he argues. It’s that business models built on selling tools plus manpower face a shift.“AI tools like Claude Cowork don’t magically replace large SaaS players immediately,” Lahoti said. “What they replace are inefficient layers, repetitive workflows, businesses that build scale on human effort rather than differentiated thinking.”He draws parallels to marketing’s evolution over the past decade. Tools didn’t kill advertising agencies, agencies that refused to adapt killed themselves. The winners focused on strategy and creativity while letting technology handle execution. “AI won’t take away all jobs,” Lahoti said. “It will take away complacency.”Umair Mohammed, founder-CEO of Nitro Commerce, sees deeper structural shifts. Having built both SaaS and AI systems for years, he believes this marks a fundamental break, not a temporary panic.“Agentic AI now threatens business models anchored in human-hour economics,” Mohammed said. Studies suggest 40% of work tasks are automatable. The World Economic Forum projects over 90 million roles could be displaced by 2030 if reskilling doesn’t keep pace.But Mohammed’s warning cuts both ways. The real danger isn’t AI replacing jobs, it’s organizations deploying it without ethics, governance, or transition plans. “If used responsibly, AI amplifies human intelligence,” he said. “Used recklessly, it destabilizes work, trust, and economies faster than we’re prepared to manage.”<h2>The evolution argument</h2>Venugopal Ganganna, co-founder and CIO at LS Digital, argues markets are pricing in capability shifts, not immediate death. “Market reactions often exaggerate the immediate impact,” Ganganna said. “What they’re reacting to is intelligence becoming embedded directly into workflows rather than sitting as a layer on top of software.”He doesn’t see established platforms disappearing overnight. Their roles will evolve, he argues. Value shifts from feature-heavy software to outcome-focused systems that compress decision time and reduce cognitive effort.For his marketing teams, that means faster synthesis, better first drafts, more informed judgment. The competitive edge comes from interpreting signals quickly, shaping narratives, and acting with confidence, not from managing platforms or producing volume.“The winners will integrate AI with domain expertise, governance, and human accountability,” Ganganna added. “Not treat it as a substitute for software or people.”<h2>What markets are really saying</h2>Anthropic’s documentation for its legal plugin explicitly states: “All outputs should be reviewed by licensed attorneys.” The company frames itself as assistant, not replacement.Markets spent Monday pricing in a different scenario, one where AI providers don’t stop at selling infrastructure but capture application revenue directly. The numbers support investor concern. Claude Code reportedly reached nearly $1 billion in annual recurring revenue within months of launch. Anthropic is preparing a $20 billion fundraise at a reported $350 billion valuation. The $285 billion wipeout delivered a clear message: AI is no longer just supporting software, it’s absorbing entire workflows.Lahoti frames the choice starkly: “Will AI disrupt SaaS? Absolutely. Will it replace the giants? Highly unlikely. The winners will rebuild their offerings around AI rather than bolt it on as a feature. The losers will be companies selling time, manpower, or basic automation dressed up as innovation.”Ganganna puts it differently but arrives at the same place: “This reinforces the need to design operations where AI augments thinking. The competitive edge comes from embedding AI into planning and performance interpretation while retaining human judgment over brand, ethics, and long-term equity.”The SaaSpocalypse isn’t about AI killing software. It’s about forcing every software company, and every business that uses it, to answer one question: What value do you create that autonomous intelligence cannot?For $285 billion worth of market capitalisation, investors decided Monday they don't have good answers yet.

Markets lost $285 billion in a single session after Anthropic released Claude Cowork plugins automating legal, marketing, and sales work. Goldman Sachs software basket crashed 6%, matching the worst decline in years. 

Markets had its worst software selloff in years on Monday, erasing $285 billion in market value, after Anthropic released plugins that let its Claude AI handle legal contracts, sales research, and financial analysis without human supervision. The storm hit everything from legacy software giants to Indian outsourcers. Thomson Reuters lost $8.2 billion in a single day, its steepest drop ever. Infosys fell 8%, TCS 6%, and the Nifty IT index shed RS1.9 lakh crore, dropping below RS30 lakh crore for the first time in months. What tensed investors wasn’t the technology itself. It was realizing that companies might soon pay for one AI agent instead of ten software licenses, or a hundred consulting hours. What changed on January 30 Anthropic launched Claude Cowork on January 12 as a desktop tool for non-coders. Users could tell it to draft documents, analyze spreadsheets, or organize files. Handy, but not revolutionary. Then came the plugins. On January 30, the company open-sourced 11 specialized add-ons on GitHub: contract reviews, NDA screening, compliance checks, legal briefings, sales prospecting, CRM integration, financial modeling, marketing automation. The AI could now execute multi-step workflows autonomously, plan tasks, spawn sub-processes, and deliver results. “When we released Claude Code, we expected developers to use it for coding,” Anthropic explained in a blog post. “They did, and then quickly began using it for almost everything else. This prompted us to build Cowork: a simpler way for anyone, not just developers, to work with Claude in the very same way.” That “same way” is the problem for software companies. Claude doesn’t just assist anymore. It executes. The billable-hour apocalypse Traders coined a new term Monday: “SaaSpocalypse.” Goldman Sachs’ software basket plunged 6%, matching its worst day since April’s tariff selloff. The Nasdaq 100 dropped 2.4% before recovering to close 1.6% down. Financial services stocks tumbled 7%. Indian IT services took the hardest hit. These companies built empires on delivering skilled labor for tasks AI can now automate. HCL Technologies and LTIMindtree fell 5-7%, Wipro dropped 4%. The Nifty IT index posted its worst month since October 2008, down 15% in January alone. Panic or prophecy? Industry experts watching the carnage split on whether markets overreacted or saw the future clearly. Adv. Varun Singh, founder and managing partner at Foresight Law Offices India, believes the selloff reflects shifting value perception rather than immediate extinction. “The way the market’s reacted demonstrates the speed with which AI programs are altering people’s ideas of what software should cost,” Varun said. “The substantial decline suggests concern that AI can consolidate various SaaS roles into one smart unit, leading to fewer tools being necessary to do specialized tasks.” But he sees evolution, not apocalypse. “This doesn’t mean big, established SaaS firms will suddenly become obsolete. The core systems remain deeply implemented and critical to business success across verticals. The actual disruption will materialize in value distribution, with businesses favoring platforms that combine legal, operational, and strategic intelligence. Consequently, incumbents will be forced to either evolve or risk margin compression rather than outright extinction.” Suyash Lahoti, partner at Wit & Chai group, believes investors missed the point. The threat isn’t software becoming irrelevant overnight, he argues. It’s that business models built on selling tools plus manpower face a shift. “AI tools like Claude Cowork don’t magically replace large SaaS players immediately,” Lahoti said. “What they replace are inefficient layers, repetitive workflows, businesses that build scale on human effort rather than differentiated thinking.” He draws parallels to marketing’s evolution over the past decade. Tools didn’t kill advertising agencies, agencies that refused to adapt killed themselves. The winners focused on strategy and creativity while letting technology handle execution. “AI won’t take away all jobs,” Lahoti said. “It will take away complacency.” Umair Mohammed, founder-CEO of Nitro Commerce, sees deeper structural shifts. Having built both SaaS and AI systems for years, he believes this marks a fundamental break, not a temporary panic. “Agentic AI now threatens business models anchored in human-hour economics,” Mohammed said. Studies suggest 40% of work tasks are automatable. The World Economic Forum projects over 90 million roles could be displaced by 2030 if reskilling doesn’t keep pace. But Mohammed’s warning cuts both ways. The real danger isn’t AI replacing jobs, it’s organizations deploying it without ethics, governance, or transition plans. “If used responsibly, AI amplifies human intelligence,” he said. “Used recklessly, it destabilizes work, trust, and economies faster than we’re prepared to manage.” The evolution argument Venugopal Ganganna, co-founder and CIO at LS Digital, argues markets are pricing in capability shifts, not immediate death. “Market reactions often exaggerate the immediate impact,” Ganganna said. “What they’re reacting to is intelligence becoming embedded directly into workflows rather than sitting as a layer on top of software.” He doesn’t see established platforms disappearing overnight. Their roles will evolve, he argues. Value shifts from feature-heavy software to outcome-focused systems that compress decision time and reduce cognitive effort. For his marketing teams, that means faster synthesis, better first drafts, more informed judgment. The competitive edge comes from interpreting signals quickly, shaping narratives, and acting with confidence, not from managing platforms or producing volume. “The winners will integrate AI with domain expertise, governance, and human accountability,” Ganganna added. “Not treat it as a substitute for software or people.” What markets are really saying Anthropic’s documentation for its legal plugin explicitly states: “All outputs should be reviewed by licensed attorneys.” The company frames itself as assistant, not replacement. Markets spent Monday pricing in a different scenario, one where AI providers don’t stop at selling infrastructure but capture application revenue directly. The numbers support investor concern. Claude Code reportedly reached nearly $1 billion in annual recurring revenue within months of launch. Anthropic is preparing a $20 billion fundraise at a reported $350 billion valuation. The $285 billion wipeout delivered a clear message: AI is no longer just supporting software, it’s absorbing entire workflows. Lahoti frames the choice starkly: “Will AI disrupt SaaS? Absolutely. Will it replace the giants? Highly unlikely. The winners will rebuild their offerings around AI rather than bolt it on as a feature. The losers will be companies selling time, manpower, or basic automation dressed up as innovation.” Ganganna puts it differently but arrives at the same place: “This reinforces the need to design operations where AI augments thinking. The competitive edge comes from embedding AI into planning and performance interpretation while retaining human judgment over brand, ethics, and long-term equity.” The SaaSpocalypse isn’t about AI killing software. It’s about forcing every software company, and every business that uses it, to answer one question: What value do you create that autonomous intelligence cannot? For $285 billion worth of market capitalisation, investors decided Monday they don't have good answers yet.

Markets lost $285 billion in a single session after Anthropic released Claude Cowork plugins automating legal, marketing, and sales work. Goldman Sachs software basket crashed 6%, matching the worst decline in years.&nbsp;

Software’s existential moment: $285B lost to Cowork threat

Google, Microsoft and Anthropic have launched new artificial intelligence tools for teachers this week, signalling a growing race among tech giants to shape how AI is used in classrooms. The announcements come at a time when many educators remain cautious about the impact of AI on learning.These companies unveiled separate initiatives aimed at teachers, students and schools, promising better access to AI tools, training and support. However, a new survey suggests that concerns around student dependence, critical thinking and data privacy remain strong among faculty members.<h2>Anthropic partners with Teach For All</h2>Anthropic announced a partnership with non-profit organisation Teach For All to bring AI tools and training to more than 100,000 educators across 63 countries, reaching over 1.5 million students. The initiative, called the AI Literacy &amp; Creator Collective, will give teachers access to Anthropic’s AI assistant, Claude.Unlike many other AI programmes, Anthropic said teachers will help shape how the technology develops. Educators will act as “co-architects”, offering feedback on how AI should be designed and used in classrooms.According to Wendy Kopp, CEO of Teach For All, teachers must play a central role if AI is to make education more fair and effective. She said educators should guide how AI tools are built, rather than simply adapting to them.&nbsp;

Anthropic partners with Teach For All

&nbsp;<h2>Google and Microsoft expand education AI</h2>Google introduced its biggest education-focused AI push so far at the Bett UK 2026 conference. The company announced free SAT practice exams through its Gemini assistant, with content reviewed by The Princeton Review. Google also announced that Khan Academy’s Writing Coach will now run on Gemini, offering writing feedback to students from Classes 5 to 12.&nbsp;In addition, Google said Gemini will be available across its Workspace for Education tools, including Gmail, Docs, Slides and Sheets, at no extra cost to schools.&nbsp;Meanwhile, Microsoft launched its Elevate for Educators programme. The initiative offers free professional training, AI-focused certifications developed with ISTE+ASCD, and access to educator networks in more than 13 languages. Microsoft said the programme supports its goal of helping 20 million people gain AI skills over the next two years.<h2>Teachers remain cautious about AI</h2>Despite the push from tech companies, many educators remain uneasy. A survey of 1,057 faculty members, released by the American Association of Colleges and Universities and Elon University, found that 95 percent believe AI will increase student overreliance on technology. Around 90 percent said it could weaken critical thinking skills.&nbsp;The survey also showed that 68 percent of faculty feel their institutions have not prepared them properly to use AI in teaching, while nearly one in four said they do not use AI tools at all.&nbsp;Eddie Watson, vice president for digital innovation at AAC&amp;U, said the findings show higher education is at a turning point, with strong warnings from faculty that AI may harm learning if not used carefully.&nbsp;<h2>Privacy and data concerns</h2>Privacy experts have also raised concerns as AI adoption in schools accelerates. While companies must follow student data protection laws such as FERPA in the United States, enforcement has often been weak.&nbsp;Some experts warn that student data protections may fade after graduation, while others suggest companies may be focused on building long-term user loyalty.Tam Winc, CEO of Secur, said that while many AI firms want to improve education, schools and parents must prioritise student safety and well-being as competition for classroom adoption increases. &nbsp;&nbsp;The growing interest in AI-powered education is also creating opportunities for specialised platforms that focus on the day-to-day needs of schools. Alongside global technology companies, India-based Neev Learn is among the platforms integrating AI into school management to support teachers, administrators, students and parents. By combining school management software with AI-powered features, the platform aims to streamline routine administrative tasks, improve communication and help educators spend more time on teaching than paperwork. As schools continue to evaluate how AI can be used responsibly, such purpose-built platforms reflect a broader shift towards practical AI applications that complement teaching rather than replace it.

Google, Microsoft and Anthropic have launched new AI tools for teachers, expanding AI use in classrooms worldwide. While firms promise support and training, surveys show educators remain worried about student dependence, weak critical thinking and data privacy risks.

Google, Microsoft and Anthropic have launched new artificial intelligence tools for teachers this week, signalling a growing race among tech giants to shape how AI is used in classrooms. The announcements come at a time when many educators remain cautious about the impact of AI on learning. These companies unveiled separate initiatives aimed at teachers, students and schools, promising better access to AI tools, training and support. However, a new survey suggests that concerns around student dependence, critical thinking and data privacy remain strong among faculty members. Anthropic partners with Teach For All Anthropic announced a partnership with non-profit organisation Teach For All to bring AI tools and training to more than 100,000 educators across 63 countries, reaching over 1.5 million students. The initiative, called the AI Literacy & Creator Collective, will give teachers access to Anthropic’s AI assistant, Claude. Unlike many other AI programmes, Anthropic said teachers will help shape how the technology develops. Educators will act as “co-architects”, offering feedback on how AI should be designed and used in classrooms. According to Wendy Kopp, CEO of Teach For All, teachers must play a central role if AI is to make education more fair and effective. She said educators should guide how AI tools are built, rather than simply adapting to them.

Google, Microsoft and Anthropic have launched new AI tools for teachers, expanding AI use in classrooms worldwide. While firms promise support and training, surveys show educators remain worried about student dependence, weak critical thinking and data privacy risks.

‘AI-powered teaching’: Google, Microsoft and Anthropic roll out AI tools for teachers

AI giant Anthropic will pay $1.5 billion to settle a US lawsuit for using pirated books to train its artificial intelligence&nbsp;models. It is the first such ruling in the AI era, which can set the tone for how businesses can offset expected losses incurred because of changing user behavior due to artificial intelligence.&nbsp;Authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson had filed a class-action lawsuit accusing Anthropic of illegally copying their books to train the firm's AI chatbot, Claude. The AI chatbot is the rival of ChatGPT.&nbsp;In June, US District Court Judge William Alsup ruled that the company's training of its Claude AI models with books -- whether bought or pirated -- transformed the works and could be deemed fair use under the law.&nbsp;Alsup compared AI training to how humans learn by reading books, saying the technology was among the most transformative.&nbsp;However, the judge rejected a blanket protection request urged by Anthropic. He said that the company's practice of downloading millions of pirated books to build a permanent digital library wasn't fair use.&nbsp;Anthropic, which is based in San Francisco, has raised $13 billion in funding, which scaled up the firm's valuation to $183 billion.&nbsp;It competes with Google, OpenAI, Meta, and Microsoft in the AI race.&nbsp;The settlement covered 500000 books at the rate of $3,000 per work. The agreement says that Anthropic will destroy the original pirated files and any copies made.<h2>What the author's guild said</h2>"This settlement sends a strong message to the AI industry that there are serious consequences when they pirate authors' works to train their AI, robbing those least able to afford it," said Mary Rasenberger, CEO of the Authors Guild, in a statement.

AI giant Anthropic will pay $1.5 billion to settle a US lawsuit for using pirated books to train its artificial intelligence models. 

AI giant Anthropic will pay $1.5 billion to settle a US lawsuit for using pirated books to train its artificial intelligence models. It is the first such ruling in the AI era, which can set the tone for how businesses can offset expected losses incurred because of changing user behavior due to artificial intelligence. Authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson had filed a class-action lawsuit accusing Anthropic of illegally copying their books to train the firm's AI chatbot, Claude. The AI chatbot is the rival of ChatGPT. In June, US District Court Judge William Alsup ruled that the company's training of its Claude AI models with books -- whether bought or pirated -- transformed the works and could be deemed fair use under the law. Alsup compared AI training to how humans learn by reading books, saying the technology was among the most transformative. However, the judge rejected a blanket protection request urged by Anthropic. He said that the company's practice of downloading millions of pirated books to build a permanent digital library wasn't fair use. Anthropic, which is based in San Francisco, has raised $13 billion in funding, which scaled up the firm's valuation to $183 billion. It competes with Google, OpenAI, Meta, and Microsoft in the AI race. The settlement covered 500000 books at the rate of $3,000 per work. The agreement says that Anthropic will destroy the original pirated files and any copies made. What the author's guild said "This settlement sends a strong message to the AI industry that there are serious consequences when they pirate authors' works to train their AI, robbing those least able to afford it," said Mary Rasenberger, CEO of the Authors Guild, in a statement.

Anthropic lawsuit: The San Francisco-based company's settlement will cover 500,000 books, which translates to $3000 per book. Mary Rasenberger, CEO of the Authors Guild, called it a major win.&nbsp;

Anthropic to pay $1.5 billion to settle lawsuit for using pirated books

Anthropic trains AI agents to audit powerful models for safetyAs artificial intelligence continues to grow more advanced, so do concerns about its safety. Anthropic, which is the AI research company behind Claude, has developed a new solution using AI to audit AI. The company has created a group of digital agents designed to investigate and fix hidden problems in large AI models before they can cause harm.&nbsp;These AI agents are now being used to test Anthropic’s own models, including Opus 4, and are seen as part of a wider move to build trust and accountability in AI development.&nbsp;<h2>Anthropic’s safety system involves three different AI agents, each with a unique task.</h2>The Investigator Agent looks deep into the model’s behaviour, acting like a digital detective. It checks the model’s training and logic to find the root cause of any suspicious patterns. For example, it can look inside the model’s neural network to understand how it’s thinking.&nbsp;Next is the Evaluation Agent, which runs controlled tests on specific issues. If a model responds in a concerning way, this agent tests how often it behaves like that, gathering clear data.&nbsp;Finally, the Red-Teaming Agent interacts with the model in many different ways, trying to trigger unexpected or dangerous responses. Any worrying behaviour it finds is flagged for review by human researchers.&nbsp;Together, these models works like a safety net scanning, testing, and reporting on how these powerful AI systems behave.&nbsp;<h2>What the tests showed</h2>To test the system, Anthropic trained a model with a hidden flaw. It had been taught to act like a sycophant, agreeing with false information to gain approval. The Investigator Agent only found this hidden goal in 13 per cent of cases when working alone.&nbsp;But when ten Investigator Agents worked together, the success rate jumped to 42 per cent. This suggests that, like in human teams, collaboration improves outcomes.&nbsp;The Evaluation Agent spotted problems in 88 per cent of test cases, while the Red-Teaming Agent uncovered 7 out of 10 intentionally hidden issues. This shows strong promise, even though the agents are not yet perfect.&nbsp;<h2>From lab to real-world models</h2>Anthropic has already added these safety agents to its live AI models. They have helped detect risks such as “prefill attacks”, where users start a sentence in a way that leads the AI into generating harmful responses.&nbsp;One of the most serious findings came from the Investigator Agent. It discovered a neural pathway inside the Opus 4 model that was linked to spreading misinformation. When this part was activated, the model ignored safety filters and created a fake news article claiming vaccines cause autism, a false claim that has been widely debunked.This finding highlighted a key issue: tools built to protect us could be misused if not properly controlled.&nbsp;<h2>A shift in AI safety work</h2>The Anthropic team admits these agents are still learning. They can miss subtle issues or get stuck on a single idea. However, they reduce the burden on human teams and allow experts to focus on higher-level decisions.&nbsp;Instead of reviewing each model manually, human researchers now design the agents and guide their efforts. This shift could help scale safety efforts as AI systems grow more powerful and complex.&nbsp;Anthropic’s approach shows how we might use AI to police AI using the same power that creates risks to help reduce them.

Anthropic using AI agents to audit its AI models like Opus 4 for hidden flaws such as misinformation bias and malicious responses. With Investigator, Evaluation and Red-Teaming agents collaborating, the system catches risks earlier. Read more below.

Anthropic trains AI agents to audit powerful models for safety As artificial intelligence continues to grow more advanced, so do concerns about its safety. Anthropic, which is the AI research company behind Claude, has developed a new solution using AI to audit AI. The company has created a group of digital agents designed to investigate and fix hidden problems in large AI models before they can cause harm. These AI agents are now being used to test Anthropic’s own models, including Opus 4, and are seen as part of a wider move to build trust and accountability in AI development. Anthropic’s safety system involves three different AI agents, each with a unique task. The Investigator Agent looks deep into the model’s behaviour, acting like a digital detective. It checks the model’s training and logic to find the root cause of any suspicious patterns. For example, it can look inside the model’s neural network to understand how it’s thinking. Next is the Evaluation Agent, which runs controlled tests on specific issues. If a model responds in a concerning way, this agent tests how often it behaves like that, gathering clear data. Finally, the Red-Teaming Agent interacts with the model in many different ways, trying to trigger unexpected or dangerous responses. Any worrying behaviour it finds is flagged for review by human researchers. Together, these models works like a safety net scanning, testing, and reporting on how these powerful AI systems behave. What the tests showed To test the system, Anthropic trained a model with a hidden flaw. It had been taught to act like a sycophant, agreeing with false information to gain approval. The Investigator Agent only found this hidden goal in 13 per cent of cases when working alone. But when ten Investigator Agents worked together, the success rate jumped to 42 per cent. This suggests that, like in human teams, collaboration improves outcomes. The Evaluation Agent spotted problems in 88 per cent of test cases, while the Red-Teaming Agent uncovered 7 out of 10 intentionally hidden issues. This shows strong promise, even though the agents are not yet perfect. From lab to real-world models Anthropic has already added these safety agents to its live AI models. They have helped detect risks such as “prefill attacks”, where users start a sentence in a way that leads the AI into generating harmful responses. One of the most serious findings came from the Investigator Agent. It discovered a neural pathway inside the Opus 4 model that was linked to spreading misinformation. When this part was activated, the model ignored safety filters and created a fake news article claiming vaccines cause autism, a false claim that has been widely debunked. This finding highlighted a key issue: tools built to protect us could be misused if not properly controlled. A shift in AI safety work The Anthropic team admits these agents are still learning. They can miss subtle issues or get stuck on a single idea. However, they reduce the burden on human teams and allow experts to focus on higher-level decisions. Instead of reviewing each model manually, human researchers now design the agents and guide their efforts. This shift could help scale safety efforts as AI systems grow more powerful and complex. Anthropic’s approach shows how we might use AI to police AI using the same power that creates risks to help reduce them.

Anthropic using AI agents to audit its AI models like Opus 4 for hidden flaws such as misinformation bias and malicious responses. With Investigator, Evaluation and Red-Teaming agents collaborating, the system catches risks earlier. Read more below.

Anthropic using AI-powered agents to detect hidden flaws in AI models

Touted as a landmark move for the artificial intelligence sector, leading AI firms OpenAI and Anthropic have entered into agreements with the United States government aimed at advancing research, testing, and evaluation of artificial intelligence (AI) technologies, according to a report by Reuters. This collaboration underscores a growing recognition of the need for safe and ethical AI development, particularly as these technologies become increasingly integrated into various aspects of society.&nbsp;

The agreements, announced on August 29, 2024, represent a significant step forward in the relationship between the government and private artificial intelligence companies. As artificial intelligence technologies evolve rapidly, the need for rigorous testing and evaluation frameworks has become paramount. The partnerships will enable OpenAI and Anthropic to work closely with the US Artificial Intelligence Safety Institute, focusing on the assessment of AI capabilities and the identification of potential safety risks.&nbsp;

Brief description of Open AI and Anthropic&nbsp;

OpenAI is an American artificial intelligence (AI) research organisation founded in December 2015 and headquartered in San Francisco, California. Its mission is to develop "safe and beneficial" artificial general intelligence (AGI), which it defines as "highly autonomous systems that outperform humans at most economically valuable work".&nbsp;

As a leading organisation in the ongoing AI boom, OpenAI is known for the GPT family of large language models, the DALL-E series of text-to-image models, and a text-to-video model named Sora. Its release of ChatGPT in November 2022 has been credited with catalysing widespread interest in generative AI.&nbsp;

Further, Anthropic PBC is a US-based artificial intelligence (AI) startup public-benefit company founded in 2021. It researches and develops AI to "study their safety properties at the technological frontier" and uses this research to deploy safe, reliable models for the public. Anthropic has developed a family of large language models (LLMs) named Claude as a competitor to OpenAI's ChatGPT and Google's Gemini.

What are the key objectives of the partnership of these firms with the US government?&nbsp;

• Enhanced safety protocols: The primary goal of these partnerships is to establish robust safety protocols for AI systems. By collaborating with government agencies, both companies aim to create a framework that ensures AI technologies are developed and deployed responsibly.&nbsp;

• Research and development: The agreements facilitate joint research initiatives that will explore various aspects of AI, including its societal impacts and ethical considerations. This research is crucial for informing future regulatory frameworks and guiding best practices in AI deployment.&nbsp;

• Public trust and transparency: By engaging with government entities, OpenAI and Anthropic are taking proactive steps to build public trust in AI technologies. Transparency in AI development is essential for alleviating concerns about misuse and unintended consequences.&nbsp;

Navigating regulatory scrutiny&nbsp;

The timing of these agreements is particularly significant as both OpenAI and Anthropic face increasing regulatory scrutiny over the safe and ethical use of AI technologies. As governments worldwide grapple with the implications of AI, these partnerships position the companies as leaders in advocating for responsible AI practices.&nbsp;

What are the implications of this partnership for the AI industry?&nbsp;

The collaboration between OpenAI, Anthropic, and the US government could set a precedent for future partnerships in the AI sector. As the industry continues to grow, the establishment of clear safety standards and ethical guidelines will be critical for fostering innovation while mitigating risks.

Touted as a landmark move for the artificial intelligence sector, leading AI firms OpenAI and Anthropic have entered into agreements with the United States government aimed at advancing research, testing, and evaluation of artificial intelligence (AI) technologies, according to a report by Reuters.

Touted as a landmark move for the artificial intelligence sector, leading AI firms OpenAI and Anthropic have entered into agreements with the United States government aimed at advancing research, testing, and evaluation of artificial intelligence (AI) technologies, according to a report by Reuters. This collaboration underscores a growing recognition of the need for safe and ethical AI development, particularly as these technologies become increasingly integrated into various aspects of society. The agreements, announced on August 29, 2024, represent a significant step forward in the relationship between the government and private artificial intelligence companies. As artificial intelligence technologies evolve rapidly, the need for rigorous testing and evaluation frameworks has become paramount. The partnerships will enable OpenAI and Anthropic to work closely with the US Artificial Intelligence Safety Institute, focusing on the assessment of AI capabilities and the identification of potential safety risks. Brief description of Open AI and Anthropic OpenAI is an American artificial intelligence (AI) research organisation founded in December 2015 and headquartered in San Francisco, California. Its mission is to develop "safe and beneficial" artificial general intelligence (AGI), which it defines as "highly autonomous systems that outperform humans at most economically valuable work". As a leading organisation in the ongoing AI boom, OpenAI is known for the GPT family of large language models, the DALL-E series of text-to-image models, and a text-to-video model named Sora. Its release of ChatGPT in November 2022 has been credited with catalysing widespread interest in generative AI. Further, Anthropic PBC is a US-based artificial intelligence (AI) startup public-benefit company founded in 2021. It researches and develops AI to "study their safety properties at the technological frontier" and uses this research to deploy safe, reliable models for the public. Anthropic has developed a family of large language models (LLMs) named Claude as a competitor to OpenAI's ChatGPT and Google's Gemini. What are the key objectives of the partnership of these firms with the US government? • Enhanced safety protocols: The primary goal of these partnerships is to establish robust safety protocols for AI systems. By collaborating with government agencies, both companies aim to create a framework that ensures AI technologies are developed and deployed responsibly. • Research and development: The agreements facilitate joint research initiatives that will explore various aspects of AI, including its societal impacts and ethical considerations. This research is crucial for informing future regulatory frameworks and guiding best practices in AI deployment. • Public trust and transparency: By engaging with government entities, OpenAI and Anthropic are taking proactive steps to build public trust in AI technologies. Transparency in AI development is essential for alleviating concerns about misuse and unintended consequences. Navigating regulatory scrutiny The timing of these agreements is particularly significant as both OpenAI and Anthropic face increasing regulatory scrutiny over the safe and ethical use of AI technologies. As governments worldwide grapple with the implications of AI, these partnerships position the companies as leaders in advocating for responsible AI practices. What are the implications of this partnership for the AI industry? The collaboration between OpenAI, Anthropic, and the US government could set a precedent for future partnerships in the AI sector. As the industry continues to grow, the establishment of clear safety standards and ethical guidelines will be critical for fostering innovation while mitigating risks.

OpenAI and Anthropic partner with US government for cutting-edge AI research

Anthropic, which is an artificial intelligence (AI) company backed by Amazon, faced a lawsuit for copyright infringement by notable authors. As reported by Reuters, Claude,the company’s AI chatbot has been accused of being trained with pirated works of writers/journalists.

Authors sue Amazon-backed Anthropic for copyright breach

anthropic

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

anthropic

Anthropic Sues US Government Over Military Ban; Dispute Escalates

Trump Urges Govt to Drop Anthropic's AI | Anthropic Sues Pentagon Over AI Blacklist

Tech Firm Anthropic Sues Donald Trump Administration Over Pentagon Blacklist

This is f***ing insane!: AI’s K-shaped future could decimate middle classes

Anthropic Vows Court Battle After Pentagon Flags AI Firm As Security Risk | WION

Anthropic Vs Pentagon: Anthropic to Challenge Pentagon “Supply Chain Risk” Label

Trump Announces Six-Month AI Phase-Out: Legal Doubts Cloud Supply Chain Ban

AI: Anthropic Defies Pentagon, Refuses Unconditional US Military AI Access | WION

Pentagon Has Issued a Sharp Ultimatum to AI Firm Anthropic on Military Use of Its Models

Claude Misuse Allegations: Fake Accounts Used To Tap AI Capabilities

What Sam Altman said on his awkward exchange with Dario Amodei on stage

WSJ: Anthropic's Claude AI Deployed in Maduro Raid

How Anthropic Surged Ahead in the Race for AI Dominance

How Anthropic Surged Ahead in the Race for AI Dominance

OpenAI and Anthropic release new AI models as rivalry intensifies

Software’s existential moment: $285B lost to Cowork threat

AI in classrooms: Google, Microsoft, and Anthropic roll out tools for teachers

Anthropic to pay $1.5 billion to settle lawsuit for using pirated books

Anthropic using AI-powered agents to detect hidden flaws in AI models

OpenAI and Anthropic partner with US government for cutting-edge AI research

Authors sue Amazon-backed Anthropic for copyright breach