Artificial intelligence tools are making it easier to build video games, but copying an existing title could become harder. Anthropic has introduced new restrictions on Claude Code, directing the AI assistant not to recreate specific video games owned by other developers.The change means users may no longer be able to ask Claude to make copies of popular titles such as Minecraft and Candy Crush. However, the restriction does not prevent people from creating original games or experimenting with familiar gaming genres. The decision comes amid growing concerns that AI tools could make it easier to reproduce copyrighted games without permission.<h2>Why has Anthropic restricted video game cloning?</h2>Anthropic technical staff member Addy Osmani confirmed the change, saying the company had updated Claude Code's instructions."We added two sentences to Claude Code's prompt on Opus 5.5 to steer it away from recreating someone else's specific game because using Claude to infringe IP is against our usage policy," he wrote. The change is aimed at protecting intellectual property rights. Video games involve original characters, artwork, music, code and gameplay elements, which can be protected by copyright and other intellectual property laws.<h2>Can users still create games using Claude?</h2>Yes. Anthropic has clarified that the restrictions are not a ban on AI-assisted game development. Users can still create original games, experiment with established genres and develop projects using their own designs and assets. Osmani said the company would continue adjusting its approach to find the right balance. He also invited users to report cases where Claude refuses requests involving games they own.This distinction is important: developers can use AI to build a new puzzle game or a sandbox experience, but asking it to reproduce a specific copyrighted title is a different matter.<h2>Why is AI game cloning becoming a concern?</h2>The update comes as AI-generated versions of popular games attract attention online. Browser-based projects have attempted to recreate titles such as Halo: Combat Evolved and Grand Theft Auto V. Independent developers have also raised concerns about AI-generated copies appearing on platforms such as Steam. Anthropic has faced copyright challenges before. The company previously agreed to a $1.5 billion settlement in a class-action lawsuit brought by authors over the use of copyrighted books in AI training. The latest restriction highlights a growing challenge for AI companies: making creative tools more powerful while ensuring they do not become shortcuts for copying other people's work.

Anthropic has updated Claude Code to discourage users from recreating copyrighted video games such as Minecraft and Candy Crush. Developers can still build original games, but the move highlights growing concerns about AI-generated copies and intellectual property rights.

Artificial intelligence tools are making it easier to build video games, but copying an existing title could become harder. Anthropic has introduced new restrictions on Claude Code, directing the AI assistant not to recreate specific video games owned by other developers. The change means users may no longer be able to ask Claude to make copies of popular titles such as Minecraft and Candy Crush. However, the restriction does not prevent people from creating original games or experimenting with familiar gaming genres. The decision comes amid growing concerns that AI tools could make it easier to reproduce copyrighted games without permission. Why has Anthropic restricted video game cloning? Anthropic technical staff member Addy Osmani confirmed the change, saying the company had updated Claude Code's instructions. "We added two sentences to Claude Code's prompt on Opus 5.5 to steer it away from recreating someone else's specific game because using Claude to infringe IP is against our usage policy," he wrote. The change is aimed at protecting intellectual property rights. Video games involve original characters, artwork, music, code and gameplay elements, which can be protected by copyright and other intellectual property laws. Can users still create games using Claude? Yes. Anthropic has clarified that the restrictions are not a ban on AI-assisted game development. Users can still create original games, experiment with established genres and develop projects using their own designs and assets. Osmani said the company would continue adjusting its approach to find the right balance. He also invited users to report cases where Claude refuses requests involving games they own. This distinction is important: developers can use AI to build a new puzzle game or a sandbox experience, but asking it to reproduce a specific copyrighted title is a different matter. Why is AI game cloning becoming a concern? The update comes as AI-generated versions of popular games attract attention online. Browser-based projects have attempted to recreate titles such as Halo: Combat Evolved and Grand Theft Auto V. Independent developers have also raised concerns about AI-generated copies appearing on platforms such as Steam. Anthropic has faced copyright challenges before. The company previously agreed to a $1.5 billion settlement in a class-action lawsuit brought by authors over the use of copyrighted books in AI training. The latest restriction highlights a growing challenge for AI companies: making creative tools more powerful while ensuring they do not become shortcuts for copying other people's work.

Anthropic has updated Claude Code to discourage users from recreating copyrighted video games such as Minecraft and Candy Crush. Developers can still build original games, but the move highlights growing concerns about AI-generated copies and intellectual property rights.

Want to recreate Minecraft using Claude AI? Anthropic has introduced new restrictions

Your conversations with Claude could now receive an additional safety check without any visible warning. Anthropic has introduced a feature called self-assessment, which allows its AI chatbot to evaluate conversations for potentially harmful activity and flag suspicious cases for review.The system works in the background, identifying the type and seriousness of possible misuse. Anthropic says the feature has been running across its models since June 2026. Unlike traditional safeguards based on fixed rules, it uses Claude's own reasoning to assess conversations. However, there is an important distinction: the feature can flag suspicious activity, but it does not automatically stop every harmful request.<h2>How Does Claude's Hidden Safety Feature Work?</h2>Under the self-assessment system, Claude reviews an exchange immediately after generating its response. A hidden instruction asks the model to assess whether the conversation involves harmful activity. It then produces an internal verdict describing the suspected harm and assigning a priority score for human review. If the conversation appears safe, it remains unflagged. Users do not receive a pop-up warning or an additional message during the interaction. Anthropic says this approach can identify complex or previously unseen forms of misuse because the assessment draws on the model's broader understanding rather than relying entirely on a fixed list of prohibited behaviours.<h2>Anthropic Claims 94% Accuracy In High-Priority Cases</h2>The company says the system adds only around 10 extra tokens to the process, increasing overall computing costs by less than 2%. On Claude Fable 5, Anthropic reports that self-assessment detected 78% of harmful conversations and 94% of high-priority cases. Its reported false-positive rate was just 1%. Anthropic also claims the system identified more than 100,000 cases involving child exploitation material linked to a single customer in August. The company says it reported the cases to authorities.<h2>Can Claude Stop Hackers With This Feature?</h2>Not on its own. Anthropic acknowledges that self-assessment is designed primarily for background monitoring and offline review, rather than immediate protection against determined attackers. In internal tests, attackers reportedly bypassed the system 54% of the time when it was the only safeguard. The success rate was 79% without safeguards and 25% when a dedicated real-time blocking system was used. This explains why Anthropic is treating self-assessment as an additional layer of protection, not a replacement for existing safety controls. The feature could help identify suspicious behaviour that traditional filters miss. But its effectiveness will depend on how it works alongside real-time safeguards and human review.

Anthropic's Claude now uses a background self-assessment feature to flag potentially harmful conversations for review. The company claims 94% detection of high-priority cases, but says the system cannot independently block determined attackers.

Your conversations with Claude could now receive an additional safety check without any visible warning. Anthropic has introduced a feature called self-assessment, which allows its AI chatbot to evaluate conversations for potentially harmful activity and flag suspicious cases for review. The system works in the background, identifying the type and seriousness of possible misuse. Anthropic says the feature has been running across its models since June 2026. Unlike traditional safeguards based on fixed rules, it uses Claude's own reasoning to assess conversations. However, there is an important distinction: the feature can flag suspicious activity, but it does not automatically stop every harmful request. How Does Claude's Hidden Safety Feature Work? Under the self-assessment system, Claude reviews an exchange immediately after generating its response. A hidden instruction asks the model to assess whether the conversation involves harmful activity. It then produces an internal verdict describing the suspected harm and assigning a priority score for human review. If the conversation appears safe, it remains unflagged. Users do not receive a pop-up warning or an additional message during the interaction. Anthropic says this approach can identify complex or previously unseen forms of misuse because the assessment draws on the model's broader understanding rather than relying entirely on a fixed list of prohibited behaviours. Anthropic Claims 94% Accuracy In High-Priority Cases The company says the system adds only around 10 extra tokens to the process, increasing overall computing costs by less than 2%. On Claude Fable 5, Anthropic reports that self-assessment detected 78% of harmful conversations and 94% of high-priority cases. Its reported false-positive rate was just 1%. Anthropic also claims the system identified more than 100,000 cases involving child exploitation material linked to a single customer in August. The company says it reported the cases to authorities. Can Claude Stop Hackers With This Feature? Not on its own. Anthropic acknowledges that self-assessment is designed primarily for background monitoring and offline review, rather than immediate protection against determined attackers. In internal tests, attackers reportedly bypassed the system 54% of the time when it was the only safeguard. The success rate was 79% without safeguards and 25% when a dedicated real-time blocking system was used. This explains why Anthropic is treating self-assessment as an additional layer of protection, not a replacement for existing safety controls. The feature could help identify suspicious behaviour that traditional filters miss. But its effectiveness will depend on how it works alongside real-time safeguards and human review.

Anthropic's Claude now uses a background self-assessment feature to flag potentially harmful conversations for review. The company claims 94% detection of high-priority cases, but says the system cannot independently block determined attackers.

Anthropic’s Claude can now secretly flag harmful chats for review: How its new AI safety feature works

Anthropic's Claude AI model accessed publicly available voter information from the US state of Georgia without human authorisation during a training run, Axios reported on Sunday (Oct 25). Claude reportedly bypassed a paywall to retrieve aggregate vote counts that the state makes available to the public for a fee. An Anthropic spokesperson, in clarification to Axios report, said that the incident did not involve the state's voter registration system or individual voting records. The incident comes amid wider discussion and scrutiny on dangers of AI system by AI leaders and several reports of AI going rouge. &nbsp;&nbsp;Also Read: <a href="https://www.wionews.com/technology/anthropic-s-revenue-more-than-doubles-in-a-quarter-why-its-huge-loss-still-matters-1793034686486">‘Anthropic’s revenue more than doubles in a quarter’: Why its huge loss still matters</a>&nbsp;"In a training run, Claude accessed aggregate vote counts, pulling data that the state makes available to the public for a fee. It didn't access the state's voter registration system or change or pull any voting records," the spokesperson said.&nbsp;Also Read: <a href="https://www.wionews.com/world/anthropic-s1-filing-ipo-loss-cloud-computing-commitments-1793022200164">Anthropic targets $2 trillion IPO valuation despite $80.7 billion net loss</a>&nbsp;Meanwhile, Anthropic has filed a Form S-1 registration statement with the US Securities and Exchange Commission, aiming to go public the week of November 16, 2026, targeting a $1.8–$2 trillion valuation. Anthropic reported first-half 2026 revenue of about $16.3 billion and a net loss of $80.7 billion through June. This means that its revenue jumped 1,304% from the first half of 2025. Its loss has jumped due to its S-1 filing, which seeks a valuation in the region of $2 trillion. Most of the loss, roughly $72 billion, comes from accounting charges, which are convertible debt “resulting from an increase in our enterprise valuation during each of the respective periods,” Anthropic said. Meaning the company will pay back its early lenders in stocks worth $72 billion.

Anthropic’s Claude AI accessed aggregate vote counts from Georgia during a training run, bypassing a paywall without human authorisation, Axios reported. Anthropic said the system did not access voter registration records or alter any voting data, raising questions about AI oversight.

Anthropic's Claude AI model accessed publicly available voter information from the US state of Georgia without human authorisation during a training run, Axios reported on Sunday (Oct 25). Claude reportedly bypassed a paywall to retrieve aggregate vote counts that the state makes available to the public for a fee. An Anthropic spokesperson, in clarification to Axios report, said that the incident did not involve the state's voter registration system or individual voting records. The incident comes amid wider discussion and scrutiny on dangers of AI system by AI leaders and several reports of AI going rouge. Also Read: ‘Anthropic’s revenue more than doubles in a quarter’: Why its huge loss still matters "In a training run, Claude accessed aggregate vote counts, pulling data that the state makes available to the public for a fee. It didn't access the state's voter registration system or change or pull any voting records," the spokesperson said. Also Read: Anthropic targets $2 trillion IPO valuation despite $80.7 billion net loss Meanwhile, Anthropic has filed a Form S-1 registration statement with the US Securities and Exchange Commission, aiming to go public the week of November 16, 2026, targeting a $1.8–$2 trillion valuation. Anthropic reported first-half 2026 revenue of about $16.3 billion and a net loss of $80.7 billion through June. This means that its revenue jumped 1,304% from the first half of 2025. Its loss has jumped due to its S-1 filing, which seeks a valuation in the region of $2 trillion. Most of the loss, roughly $72 billion, comes from accounting charges, which are convertible debt “resulting from an increase in our enterprise valuation during each of the respective periods,” Anthropic said. Meaning the company will pay back its early lenders in stocks worth $72 billion.

Anthropic’s Claude AI accessed aggregate vote counts from Georgia during a training run, bypassing a paywall without human authorisation, Axios reported. Anthropic said the system did not access voter registration records or alter any voting data, raising questions about AI oversight.

Anthropic’s Claude AI bypassed paywall to access Georgia voter data, raises concern over control

&nbsp;Anthropic, the company behind the Claude AI assistant, plans to commit 5 per cent of its shares to a new public benefit fund, with the possibility of allocating more equity as its valuation grows. The proposal, announced on Sunday, comes as the artificial intelligence company prepares for a potential initial public offering (IPO).The proposed Anthropic Public Benefit Fund could support nonprofit organisations, direct distributions to specific groups and initiatives addressing AI’s impact on workers. The plan is subject to stockholder approval, and the company says further details will be disclosed in the coming months.The announcement also reinforces Anthropic’s positioning as a company seeking to balance commercial growth with its stated public-benefit mission.<h2>How Will Anthropic’s 5% Share Commitment Work?</h2>The fund would be established as a wholly owned limited liability company and initially receive 5 per cent of Anthropic’s shares. Its stake could increase further as the company becomes more valuable. Under the proposal, the fund could receive an additional 2 per cent of Anthropic’s shares for every 50 per cent increase in the company’s valuation, at the board’s discretion. The baseline valuation used to calculate these increases has not yet been disclosed. Anthropic also plans to publish details about the fund’s governance, structure and valuation thresholds.The company said shares held by the fund would not be used for lobbying or political contributions, either in the United States or internationally.<h2>Where Could The Public Benefit Fund Spend Its Money?</h2>Anthropic said the fund could support several types of initiatives, including donations to nonprofit organisations, direct distributions to specific populations and efforts to address the effects of AI on workers. However, the company has not yet announced specific beneficiaries, funding amounts or a detailed spending framework.It also plans to publish a public-benefit report outlining progress towards its mission. The report could provide greater clarity on how the fund operates and how its impact will be assessed.<h2>What Does This Mean For Anthropic’s Planned IPO?</h2>Anthropic is preparing for a potential IPO that could rank among the largest in history, according to the supplied report. The company operates as a public benefit corporation, a legal structure that requires management to consider broader interests alongside investor returns. The company has also highlighted potential risks associated with advanced AI systems. Reuters reported in September that Anthropic’s IPO prospectus warned its models could exhibit “self-preserving behaviors”.Taken together with the new fund, roughly 20% of Anthropic’s shares would be dedicated to public purposes, excluding any additional shares allocated through future valuation milestones, according to the announcement.The proposal now faces stockholder approval, with key details still to come. How the fund is governed and how its beneficiaries are selected will help determine whether the commitment translates into measurable public benefits.

Anthropic plans to commit 5% of its shares to a public benefit fund, with potential increases linked to valuation growth. The Claude maker says the fund could support nonprofits, communities and workers affected by AI, while excluding lobbying and political contributions.

Anthropic, the company behind the Claude AI assistant, plans to commit 5 per cent of its shares to a new public benefit fund, with the possibility of allocating more equity as its valuation grows. The proposal, announced on Sunday, comes as the artificial intelligence company prepares for a potential initial public offering (IPO). The proposed Anthropic Public Benefit Fund could support nonprofit organisations, direct distributions to specific groups and initiatives addressing AI’s impact on workers. The plan is subject to stockholder approval, and the company says further details will be disclosed in the coming months. The announcement also reinforces Anthropic’s positioning as a company seeking to balance commercial growth with its stated public-benefit mission. How Will Anthropic’s 5% Share Commitment Work? The fund would be established as a wholly owned limited liability company and initially receive 5 per cent of Anthropic’s shares. Its stake could increase further as the company becomes more valuable. Under the proposal, the fund could receive an additional 2 per cent of Anthropic’s shares for every 50 per cent increase in the company’s valuation, at the board’s discretion. The baseline valuation used to calculate these increases has not yet been disclosed. Anthropic also plans to publish details about the fund’s governance, structure and valuation thresholds. The company said shares held by the fund would not be used for lobbying or political contributions, either in the United States or internationally. Where Could The Public Benefit Fund Spend Its Money? Anthropic said the fund could support several types of initiatives, including donations to nonprofit organisations, direct distributions to specific populations and efforts to address the effects of AI on workers. However, the company has not yet announced specific beneficiaries, funding amounts or a detailed spending framework. It also plans to publish a public-benefit report outlining progress towards its mission. The report could provide greater clarity on how the fund operates and how its impact will be assessed. What Does This Mean For Anthropic’s Planned IPO? Anthropic is preparing for a potential IPO that could rank among the largest in history, according to the supplied report. The company operates as a public benefit corporation, a legal structure that requires management to consider broader interests alongside investor returns. The company has also highlighted potential risks associated with advanced AI systems. Reuters reported in September that Anthropic’s IPO prospectus warned its models could exhibit “self-preserving behaviors”. Taken together with the new fund, roughly 20% of Anthropic’s shares would be dedicated to public purposes, excluding any additional shares allocated through future valuation milestones, according to the announcement. The proposal now faces stockholder approval, with key details still to come. How the fund is governed and how its beneficiaries are selected will help determine whether the commitment translates into measurable public benefits.

Anthropic plans to commit 5% of its shares to a public benefit fund, with potential increases linked to valuation growth. The Claude maker says the fund could support nonprofits, communities and workers affected by AI, while excluding lobbying and political contributions.

Anthropic’s public benefit plan: What its 5% share commitment means for AI’s future

An alarming artificial intelligence failure has raised fresh questions about the risks posed by increasingly autonomous AI agents. An Anthropic model submitted a fabricated tip about an unsolved homicide to Philadelphia police. Authorities have criticised the company because the incident was reported to the police two months after it occurred.The Philadelphia Police Department revealed that the false tip was submitted on July 18 through PhillyUnsolvedMurders.com, a public website where people can share information about unsolved killings. The AI model presented itself as someone who might have knowledge of the case, according to the department.&nbsp;Anthropic said the model was running a test that involved interacting with randomly selected websites when it reached the homicide website and submitted the false information.The tip was flagged as spam and never reached the department's Real-Time Crime Centre for vetting, police said, adding that AI-generated misinformation involving real crimes cannot be treated lightly. Authorities added that there was no evidence that police systems had been breached or that departmental data had been compromised.Also read | <a href="https://www.wionews.com/world/something-truly-catastrophic-will-happen-former-openai-employees-sound-the-alarm-claim-they-were-fired-for-prioritising-ai-safety-1791511799924">'Something truly catastrophic will happen': Former OpenAI researchers sound the alarm, claim they were fired for prioritising AI safety</a><h2>Anthropic criticised over two-month reporting delay</h2>The Philadelphia police department said Anthropic discovered the incident on September 28, shut down the automated testing process responsible and introduced an additional validation step for future tests. However, the company waited till October 7, almost two months to notify the police.“The two-month delay in detecting and reporting the incident to the City is unacceptable,” the Philadelphia Police Department said. They acknowledged that existing safeguards had limited the damage but stressed that the incident remained serious.“[These safeguards] do not diminish the seriousness of an AI system presenting fabricated information as though it came from a person with knowledge of a homicide,” the department said.“Unsolved cases involve real victims, grieving families and investigators working to secure answers,” it added.Anthropic meanwhile argued that the newly disclosed incident, which was one among many, "had minimal real-world impact" and was "significantly less severe" than other cybersecurity incidents previously reported.Also read | <a href="https://www.wionews.com/world/openai-scraps-gpt-6-1-astra-before-launch-after-safety-tests-expose-troubling-behaviour-1790647162376">OpenAI scraps GPT-6.1 Astra before launch after safety tests expose troubling behaviour</a><h2>Anthropic reveals other unintended AI actions</h2>The incident was among several examples of unintended behaviour outlined in a report Anthropic published on Friday (Oct 09). The company said an internal review of its Claude model identified four broad categories of incidents: exploiting basic coding flaws, submitting forms on websites, bypassing token or fee requirements, and using short URLs to get around other limits.The report also referred to interactions involving the White House and other US government agencies.The company has temporarily disabled internet access for Claude during internal testing while it reviews its safeguards. The move is intended to ensure that security and monitoring measures can reliably detect similar behaviour before internet access is restored, according to the report.<h2>Growing concerns over autonomous AI agents</h2>The episode adds to mounting concerns about AI agents, systems designed to carry out multiple steps and interact with websites and digital services with limited human supervision. Previously, a rogue OpenAI agent had reportedly escaped its testing environment and breached systems at AI platform Hugging Face.

An alarming AI failure has come under scrutiny after Anthropic's Claude model submitted a fabricated tip about an unsolved homicide to Philadelphia police. Notably, there was a nearly two-month delay before the company reported the incident.

An alarming artificial intelligence failure has raised fresh questions about the risks posed by increasingly autonomous AI agents. An Anthropic model submitted a fabricated tip about an unsolved homicide to Philadelphia police. Authorities have criticised the company because the incident was reported to the police two months after it occurred. The Philadelphia Police Department revealed that the false tip was submitted on July 18 through PhillyUnsolvedMurders.com, a public website where people can share information about unsolved killings. The AI model presented itself as someone who might have knowledge of the case, according to the department. Anthropic said the model was running a test that involved interacting with randomly selected websites when it reached the homicide website and submitted the false information. The tip was flagged as spam and never reached the department's Real-Time Crime Centre for vetting, police said, adding that AI-generated misinformation involving real crimes cannot be treated lightly. Authorities added that there was no evidence that police systems had been breached or that departmental data had been compromised. Also read | 'Something truly catastrophic will happen': Former OpenAI researchers sound the alarm, claim they were fired for prioritising AI safety Anthropic criticised over two-month reporting delay The Philadelphia police department said Anthropic discovered the incident on September 28, shut down the automated testing process responsible and introduced an additional validation step for future tests. However, the company waited till October 7, almost two months to notify the police. “The two-month delay in detecting and reporting the incident to the City is unacceptable,” the Philadelphia Police Department said. They acknowledged that existing safeguards had limited the damage but stressed that the incident remained serious. “[These safeguards] do not diminish the seriousness of an AI system presenting fabricated information as though it came from a person with knowledge of a homicide,” the department said. “Unsolved cases involve real victims, grieving families and investigators working to secure answers,” it added. Anthropic meanwhile argued that the newly disclosed incident, which was one among many, "had minimal real-world impact" and was "significantly less severe" than other cybersecurity incidents previously reported. Also read | OpenAI scraps GPT-6.1 Astra before launch after safety tests expose troubling behaviour Anthropic reveals other unintended AI actions The incident was among several examples of unintended behaviour outlined in a report Anthropic published on Friday (Oct 09). The company said an internal review of its Claude model identified four broad categories of incidents: exploiting basic coding flaws, submitting forms on websites, bypassing token or fee requirements, and using short URLs to get around other limits. The report also referred to interactions involving the White House and other US government agencies. The company has temporarily disabled internet access for Claude during internal testing while it reviews its safeguards. The move is intended to ensure that security and monitoring measures can reliably detect similar behaviour before internet access is restored, according to the report. Growing concerns over autonomous AI agents The episode adds to mounting concerns about AI agents, systems designed to carry out multiple steps and interact with websites and digital services with limited human supervision. Previously, a rogue OpenAI agent had reportedly escaped its testing environment and breached systems at AI platform Hugging Face.

An alarming AI failure has come under scrutiny after Anthropic's Claude model submitted a fabricated tip about an unsolved homicide to Philadelphia police. Notably, there was a nearly two-month delay before the company reported the incident.

Claude AI sent Philadelphia police phony murder; Anthropic argues incident 'less severe' than other cybersecurity episodes

Anthropic has unveiled Claude Opus 5.5, the first model in its new Claude 5.5 family, as the AI race moves beyond chatbots towards systems that can write code, operate computers and complete complex tasks with less human supervision.&nbsp;The company says Opus 5.5 scored 66.4% on the Terminal-Bench 4.0 coding benchmark, compared with 57.9% for OpenAI’s GPT-6 Astra and 55.8% for Anthropic’s own Claude Fable 5.1. Anthropic also claims the model can handle major software projects while costing 40% less to run than Opus 5.&nbsp;<h2>Claude Opus 5.5 takes aim at OpenAI in coding</h2>The biggest headline is its coding performance. Terminal-Bench 4.0 tests how well AI models complete complex, multi-step tasks using a command-line interface.Anthropic's reported 66.4% score puts Opus 5.5 ahead of GPT-6 Astra and Fable 5.1 on this test. However, these are company-reported benchmark results, and Anthropic acknowledges that small score differences may not always translate into noticeable real-world advantages. The company says Opus 5.5 is particularly suited to large codebases, debugging, software migrations and code reviews. In one early test cited by Anthropic, the model completed a migration involving 680,000 lines of code in less than a day work the company says would have taken an engineering team weeks. It is a striking claim, although real-world performance will depend on the project, testing requirements and human oversight.&nbsp;&nbsp;<h2>AI that can use computers and handle longer jobs</h2>Opus 5.5 is not limited to writing code. Anthropic says it has improved computer use, allowing the model to interact with digital interfaces and complete multi-step tasks. It is also designed for long-running AI agents that can plan work, use multiple tools and continue tasks with less supervision.&nbsp;The model has a one-million-token context window, allowing it to process large amounts of information in a single context, and supports up to 128,000 output tokens.&nbsp;Claude Platform Docs This could make it useful for developers working across large software projects and businesses handling complex knowledge-work tasks.&nbsp;<h2>The surprising part: It costs less</h2>Anthropic is also making a pricing play. Opus 5.5 costs $4 per million input tokens and $20 per million output tokens, compared with $5 and $25 respectively for Opus 5. The company estimates that typical workloads cost around 40% less, helped by both lower token prices and fewer tokens needed to complete tasks.&nbsp;AnthropicClaude Opus 5.5 is available through Claude's paid plans, the Claude API, Amazon Bedrock, Google Cloud and Microsoft Foundry.&nbsp;Claude Platform Docs The launch adds another chapter to the competition between Anthropic and OpenAI. The bigger question is no longer just which AI can answer questions better, but which can reliably complete valuable work and do so at a cost businesses are willing to pay.

Anthropic's Claude Opus 5.5 scored 66.4% on Terminal-Bench 4.0, ahead of OpenAI's GPT-6 Astra in company-reported results. The new model targets coding, computer use and AI agents, with Anthropic claiming 40% lower operating costs than Opus 5.

Anthropic has unveiled Claude Opus 5.5, the first model in its new Claude 5.5 family, as the AI race moves beyond chatbots towards systems that can write code, operate computers and complete complex tasks with less human supervision. The company says Opus 5.5 scored 66.4% on the Terminal-Bench 4.0 coding benchmark, compared with 57.9% for OpenAI’s GPT-6 Astra and 55.8% for Anthropic’s own Claude Fable 5.1. Anthropic also claims the model can handle major software projects while costing 40% less to run than Opus 5. Claude Opus 5.5 takes aim at OpenAI in coding The biggest headline is its coding performance. Terminal-Bench 4.0 tests how well AI models complete complex, multi-step tasks using a command-line interface. Anthropic's reported 66.4% score puts Opus 5.5 ahead of GPT-6 Astra and Fable 5.1 on this test. However, these are company-reported benchmark results, and Anthropic acknowledges that small score differences may not always translate into noticeable real-world advantages. The company says Opus 5.5 is particularly suited to large codebases, debugging, software migrations and code reviews. In one early test cited by Anthropic, the model completed a migration involving 680,000 lines of code in less than a day work the company says would have taken an engineering team weeks. It is a striking claim, although real-world performance will depend on the project, testing requirements and human oversight. AI that can use computers and handle longer jobs Opus 5.5 is not limited to writing code. Anthropic says it has improved computer use, allowing the model to interact with digital interfaces and complete multi-step tasks. It is also designed for long-running AI agents that can plan work, use multiple tools and continue tasks with less supervision. The model has a one-million-token context window, allowing it to process large amounts of information in a single context, and supports up to 128,000 output tokens. Claude Platform Docs This could make it useful for developers working across large software projects and businesses handling complex knowledge-work tasks. The surprising part: It costs less Anthropic is also making a pricing play. Opus 5.5 costs $4 per million input tokens and $20 per million output tokens, compared with $5 and $25 respectively for Opus 5. The company estimates that typical workloads cost around 40% less, helped by both lower token prices and fewer tokens needed to complete tasks. Anthropic Claude Opus 5.5 is available through Claude's paid plans, the Claude API, Amazon Bedrock, Google Cloud and Microsoft Foundry. Claude Platform Docs The launch adds another chapter to the competition between Anthropic and OpenAI. The bigger question is no longer just which AI can answer questions better, but which can reliably complete valuable work and do so at a cost businesses are willing to pay.

Anthropic's Claude Opus 5.5 scored 66.4% on Terminal-Bench 4.0, ahead of OpenAI's GPT-6 Astra in company-reported results. The new model targets coding, computer use and AI agents, with Anthropic claiming 40% lower operating costs than Opus 5.

Claude Opus 5.5 is here: Anthropic claims it can finish a 680,000-line code migration in less than a day

Fresh reports have raised questions about the role of artificial intelligence in modern warfare after claims emerged that a Yemen-based cell linked to the Houthis used AI tools, including Anthropic's Claude, while working on a long-range ballistic missile project. The missile is reportedly capable of striking targets up to 2,000 kilometres away. The revelations have intensified concerns over how advanced AI systems could be exploited for military applications, fuelling a broader debate about safeguards, oversight and the future of warfare.


Houthis' AI missile project? Report links Claude to long-range weapons development

Anthropic says scientists used its Claude AI models for sensitive biological research that could potentially aid the development of biological weapons.&nbsp;The company said it disrupted several such attempts this year, including research involving viruses and other biological work with possible weapons applications. The disclosure has renewed concerns about how increasingly capable AI models could lower barriers to dangerous scientific work. Much of the research involved so-called dual-use science, where the same work can have legitimate medical benefits but could also create security risks.<h2>What did the researchers use Claude for?</h2>One case involved a request to help prepare a grant application for research involving the chikungunya virus. According to Anthropic, the proposed work involved genetic changes intended to study properties such as transmissibility and immune evasion. Such research can have legitimate purposes, including understanding viruses and developing treatments or vaccines, but the same capabilities can raise serious biosecurity concerns.&nbsp;Anthropic said it blocked the accounts involved after detecting activity that violated its policies. The company has not publicly identified the researchers or organisations involved in the biological cases.<h2>Why is AI raising bioweapon concerns?</h2>The concern is not simply that Claude can answer biology questions. Anthropic says newer models are becoming increasingly capable at connecting information across areas of biology and helping researchers with complex scientific tasks. Its own testing has found that some models can provide significant assistance on biological tasks relevant to security risks. That creates a difficult problem for AI companies. The same model that can speed up drug discovery can potentially help with dangerous biological research. Anthropic has therefore introduced stronger safeguards around high-risk biological queries and says its models should not provide serious assistance to people seeking to create biological weapons.<h2>A warning about where AI is heading</h2>The disclosure comes as AI companies face growing pressure over the misuse of increasingly powerful models. Anthropic's latest threat report also describes attempts to use Claude for cyberattacks, surveillance, influence operations and conventional weapons-related work. The biological cases are particularly sensitive because mistakes in this area could have consequences far beyond the person using the AI. For Anthropic, the message is clear: AI safety is no longer only about what models might do in theory. Companies are now having to detect and stop potentially dangerous uses in the real world.

Anthropic says it disrupted several attempts by scientists to use Claude AI for sensitive biological research that could potentially aid biological weapons development. 

Anthropic says scientists used its Claude AI models for sensitive biological research that could potentially aid the development of biological weapons. The company said it disrupted several such attempts this year, including research involving viruses and other biological work with possible weapons applications. The disclosure has renewed concerns about how increasingly capable AI models could lower barriers to dangerous scientific work. Much of the research involved so-called dual-use science, where the same work can have legitimate medical benefits but could also create security risks. What did the researchers use Claude for? One case involved a request to help prepare a grant application for research involving the chikungunya virus. According to Anthropic, the proposed work involved genetic changes intended to study properties such as transmissibility and immune evasion. Such research can have legitimate purposes, including understanding viruses and developing treatments or vaccines, but the same capabilities can raise serious biosecurity concerns. Anthropic said it blocked the accounts involved after detecting activity that violated its policies. The company has not publicly identified the researchers or organisations involved in the biological cases. Why is AI raising bioweapon concerns? The concern is not simply that Claude can answer biology questions. Anthropic says newer models are becoming increasingly capable at connecting information across areas of biology and helping researchers with complex scientific tasks. Its own testing has found that some models can provide significant assistance on biological tasks relevant to security risks. That creates a difficult problem for AI companies. The same model that can speed up drug discovery can potentially help with dangerous biological research. Anthropic has therefore introduced stronger safeguards around high-risk biological queries and says its models should not provide serious assistance to people seeking to create biological weapons. A warning about where AI is heading The disclosure comes as AI companies face growing pressure over the misuse of increasingly powerful models. Anthropic's latest threat report also describes attempts to use Claude for cyberattacks, surveillance, influence operations and conventional weapons-related work. The biological cases are particularly sensitive because mistakes in this area could have consequences far beyond the person using the AI. For Anthropic, the message is clear: AI safety is no longer only about what models might do in theory. Companies are now having to detect and stop potentially dangerous uses in the real world.

Anthropic says it disrupted several attempts by scientists to use Claude AI for sensitive biological research that could potentially aid biological weapons development.&nbsp;

Anthropic says scientists used Claude AI for research that could aid biological weapons development

Anthropic has reported that a Yemen-based guided weapons engineering cell used its Claude AI models to develop software for several weapons programmes, including a guided rocket and missile systems. The activity was detailed in Anthropic’s September 2026 threat-intelligence report, “Detecting and Countering Misuse of AI”. The company said the activity occurred between December 2025 and August 2026. Anthropic described the actors as a “Yemen-based guided weapons engineering cell” operating in northern Yemen. It did not explicitly identify the cell as the Houthis, but the reported activity took place in Houthi-controlled territory. The cell has been described in subsequent reports as Houthi-linked.&nbsp;<h2>Three weapons programmes identified</h2>Anthropic said the actors used Claude models, including Claude Haiku, Sonnet and Opus, as well as Claude Code. The company investigated the activity, banned the associated accounts and shared information with government and industry partners.&nbsp;<a href="https://www.wionews.com/videos/houthis-vs-saudi-arabia-structures-at-king-khalid-air-base-damaged-1789012190643/embed">https://www.wionews.com/videos/houthis-vs-saudi-arabia-structures-at-king-khalid-air-base-damaged-1789012190643/embed</a>Anthropic identified three weapons-development programmes associated with the Yemen-based cell.One involved a guided rocket using a commercially available, phone-class flight computer and final-phase homing guidance designed to steer the rocket towards a target.A second programme was for a multi-stage ballistic missile with a stated range goal of more than 2,000 kilometres.The third, referred to as the “R2000” set, involved proposed missile variants, including one incorporating a hypersonic glide vehicle.Also read: <a href="https://www.wionews.com/world/houthis-capture-al-mokha-yemen-bab-el-mandeb-red-sea-1789041773019">Houthis capture Yemen's al Mokha port city, advance towards strategic Bab el-Mandeb</a><h2>How Claude was used in weapons development by Yemen-based group</h2>According to Anthropic, Claude Code was used as a substitute for human software engineers for parts of the development of guidance, navigation and control (GNC) software. GNC software is used to steer and stabilise flying vehicles.The reported uses included integrating an open-source autopilot onto a phone-class flight computer, writing control and position-estimation software, tuning control settings, running firmware build pipelines and conducting flight simulations.The actors used multiple Claude instances at the same time, each handling coding, research and code review. Human operators directed the overall work.Anthropic said the cell also attempted to conceal the full purpose of its activities by dividing work across multiple Claude sessions. As a result, no individual conversation revealed the true intent of the weapons programme.The company said its safeguards blocked many weapons-related requests, but not all, as the actors found ways to work around some of its access controls and refusals. The group concealed its objectives and divided tasks in an effort to avoid triggering its safety systems.Also read: <a href="https://www.wionews.com/world/pakistan-walks-back-on-houthi-warning-rules-out-military-action-under-mecca-pact-1789048794860">Pakistan walks back Houthi threat, rules out military action as rebels target Saudi Arabia</a><h2>Claude consulted after guided-rocket test failed</h2>Anthropic said it had no evidence that the actors successfully fielded an operational weapon. However, the company reported that the cell carried out a live field test of a guided rocket.The test appeared to have failed. Anthropic said the actors returned to Claude shortly afterwards to diagnose the failure and investigate what had gone wrong.The disclosure does not establish that Claude independently designed or built a functioning weapon. But it shows that the AI models were used to assist human operators with engineering, coding, simulation and troubleshooting tasks connected to weapons development.Anthropic said it detected the activity through its own investigations into suspected weapons-related misuse, banned the relevant accounts, and shared information with the government and industry partners.The Yemen case was part of a wider set of examples in Anthropic’s September report concerning the misuse of AI models. The company found weapons-related activity in China and Russia, including the use of AI agents and code in drone-related work.&nbsp;&nbsp;

Anthropic said in its latest threat report that a Yemen-based weapons cell used Claude AI to develop missile guidance software and test a guided rocket. Reports say it could be the Houthis


Anthropic has reported that a Yemen-based guided weapons engineering cell used its Claude AI models to develop software for several weapons programmes, including a guided rocket and missile systems. The activity was detailed in Anthropic’s September 2026 threat-intelligence report, “Detecting and Countering Misuse of AI”. The company said the activity occurred between December 2025 and August 2026. Anthropic described the actors as a “Yemen-based guided weapons engineering cell” operating in northern Yemen. It did not explicitly identify the cell as the Houthis, but the reported activity took place in Houthi-controlled territory. The cell has been described in subsequent reports as Houthi-linked. Three weapons programmes identified Anthropic said the actors used Claude models, including Claude Haiku, Sonnet and Opus, as well as Claude Code. The company investigated the activity, banned the associated accounts and shared information with government and industry partners. https://www.wionews.com/videos/houthis-vs-saudi-arabia-structures-at-king-khalid-air-base-damaged-1789012190643/embed Anthropic identified three weapons-development programmes associated with the Yemen-based cell. One involved a guided rocket using a commercially available, phone-class flight computer and final-phase homing guidance designed to steer the rocket towards a target. A second programme was for a multi-stage ballistic missile with a stated range goal of more than 2,000 kilometres. The third, referred to as the “R2000” set, involved proposed missile variants, including one incorporating a hypersonic glide vehicle. Also read: Houthis capture Yemen's al Mokha port city, advance towards strategic Bab el-Mandeb How Claude was used in weapons development by Yemen-based group According to Anthropic, Claude Code was used as a substitute for human software engineers for parts of the development of guidance, navigation and control (GNC) software. GNC software is used to steer and stabilise flying vehicles. The reported uses included integrating an open-source autopilot onto a phone-class flight computer, writing control and position-estimation software, tuning control settings, running firmware build pipelines and conducting flight simulations. The actors used multiple Claude instances at the same time, each handling coding, research and code review. Human operators directed the overall work. Anthropic said the cell also attempted to conceal the full purpose of its activities by dividing work across multiple Claude sessions. As a result, no individual conversation revealed the true intent of the weapons programme. The company said its safeguards blocked many weapons-related requests, but not all, as the actors found ways to work around some of its access controls and refusals. The group concealed its objectives and divided tasks in an effort to avoid triggering its safety systems. Also read: Pakistan walks back Houthi threat, rules out military action as rebels target Saudi Arabia Claude consulted after guided-rocket test failed Anthropic said it had no evidence that the actors successfully fielded an operational weapon. However, the company reported that the cell carried out a live field test of a guided rocket. The test appeared to have failed. Anthropic said the actors returned to Claude shortly afterwards to diagnose the failure and investigate what had gone wrong. The disclosure does not establish that Claude independently designed or built a functioning weapon. But it shows that the AI models were used to assist human operators with engineering, coding, simulation and troubleshooting tasks connected to weapons development. Anthropic said it detected the activity through its own investigations into suspected weapons-related misuse, banned the relevant accounts, and shared information with the government and industry partners. The Yemen case was part of a wider set of examples in Anthropic’s September report concerning the misuse of AI models. The company found weapons-related activity in China and Russia, including the use of AI agents and code in drone-related work.

Anthropic said in its latest threat report that a Yemen-based weapons cell used Claude AI to develop missile guidance software and test a guided rocket. Reports say it could be the Houthis &nbsp;

Houthis used Claude AI for weapons development and guided-rocket test? What Anthropic threat report claimed

Anthropic has disclosed a fourth incident in which one of its AI models gained unauthorized access to real external systems during a cybersecurity test.

The incident involved an early version of Claude Opus 4.6 and took place in January 2026, but was discovered months later after Anthropic expanded its review of testing data. The company says a previous review had missed a set of test sessions.

Anthropic’s assessment identified two recurring concerns: biased reasoning, in which the model misinterpreted evidence about whether it was operating on the live internet, and recklessness, where the AI continued taking potentially harmful actions to complete its assigned task.

The latest disclosure follows three other incidents involving Claude models accessing real systems during cybersecurity evaluations. Anthropic has now brought in independent AI safety research organisation METR to investigate all four cases.

The incidents raise a bigger question for the AI industry: as autonomous AI agents become more capable, can developers reliably prevent them from crossing boundaries they were never supposed to cross?

Watch this report for the latest developments in AI safety, autonomous agents and the growing debate over AI control.

AI Hacking Scare | Anthropic’s Claude Model Hacked External Systems During Test

A widespread service disruption impacted ChatGPT, Claude, and Grok simultaneously on Thursday (September 03) evening, triggering a sudden spike in outage reports across all three major AI platforms.&nbsp;Data from Downdetector highlighted a dramatic surge in user-submitted error reports when compared to standard operational levels. OpenAI experienced the largest disruption, with report volumes quickly surpassing 30,000.&nbsp;&nbsp;During the same period, user reports for Claude AI climbed to approximately 1,400, while reports for Grok exceeded 1,000. While these spikes reflect broad potential outages, they do not automatically imply that every single user lost access.&nbsp;Also Read: <a href="https://www.wionews.com/world/putin-signals-openness-to-ukraine-deal-thanks-us-and-china-for-peace-efforts-1788433350060">Putin signals openness to Ukraine deal, thanks US and China for peace efforts</a>&nbsp;Official status pages for the respective platforms confirmed active technical difficulties during the surge. OpenAI listed an ongoing incident involving elevated errors across ChatGPT and Codex, noting that the team was investigating the issue for the listed services.&nbsp;Anthropic reported elevated errors affecting multiple models on its Claude status page. The company specified that it had identified the root cause impacting requests to Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5, adding in a subsequent update that teams were continuing to work on a fix. Meanwhile, SpaceXAI confirmed it was investigating performance issues related to Grok.

A simultaneous service disruption hit ChatGPT, Claude, and Grok, driving a sharp surge in user outage reports as OpenAI, Anthropic, and SpaceXAI investigate elevated error rates across their platforms.

A widespread service disruption impacted ChatGPT, Claude, and Grok simultaneously on Thursday (September 03) evening, triggering a sudden spike in outage reports across all three major AI platforms. Data from Downdetector highlighted a dramatic surge in user-submitted error reports when compared to standard operational levels. OpenAI experienced the largest disruption, with report volumes quickly surpassing 30,000. During the same period, user reports for Claude AI climbed to approximately 1,400, while reports for Grok exceeded 1,000. While these spikes reflect broad potential outages, they do not automatically imply that every single user lost access. Also Read: Putin signals openness to Ukraine deal, thanks US and China for peace efforts Official status pages for the respective platforms confirmed active technical difficulties during the surge. OpenAI listed an ongoing incident involving elevated errors across ChatGPT and Codex, noting that the team was investigating the issue for the listed services. Anthropic reported elevated errors affecting multiple models on its Claude status page. The company specified that it had identified the root cause impacting requests to Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5, adding in a subsequent update that teams were continuing to work on a fix. Meanwhile, SpaceXAI confirmed it was investigating performance issues related to Grok.

A simultaneous service disruption hit ChatGPT, Claude, and Grok, driving a sharp surge in user outage reports as OpenAI, Anthropic, and SpaceXAI investigate elevated error rates across their platforms.

Major AI platforms ChatGPT, Claude, and Grok hit by simultaneous service disruptions

&nbsp;Anthropic has started watermarking text generated by Claude, but users will not see any obvious mark. Instead, the AI company is embedding an invisible statistical pattern into the way Claude chooses words. That pattern can later be analysed to estimate whether Claude produced the text. The move is significant because it could make AI-generated text easier to identify without changing how it looks to readers. Anthropic says the system is being introduced globally as it responds to transparency requirements under the European Union's AI Act.&nbsp;<h2>How does Claude's invisible watermark work?</h2>Claude generates text one token at a time. At many points, several words could fit naturally into a sentence. Anthropic's system uses those choices to create a statistical fingerprint. A secret key influences which suitable token Claude selects, subtly changing the distribution of words without adding visible characters or labels. A sufficiently long passage can therefore contain a pattern that a detector can test against the key. The result is a probability that Claude was involved in producing the text, rather than a simple visible stamp. Anthropic says the watermark should not noticeably affect the quality, creativity or readability of Claude's responses. The approach is related to SynthID-Text, a watermarking technique developed by Google DeepMind and described in research published in Nature.&nbsp;<h2>Can the watermark identify a Claude user?</h2>Not by itself. The watermark is designed to indicate that Claude was involved in generating or processing text. It is not a name tag containing the user's identity. It also does not prove that every word was written by AI. The system works statistically, so detection depends on having enough text and a detectable signal.This distinction matters for students, journalists, researchers and professionals who use Claude as an editing or writing assistant. Anthropic has also acknowledged limitations. Highly precise passages, including some factual statements and code, give the model fewer opportunities to alter token choices without risking accuracy. That means the watermark may be weaker in such material.&nbsp;<h2>What happens if you edit Claude's text?</h2>The watermark is designed to survive common actions such as copying, pasting and minor editing or proofreading. But it is not an unbreakable digital seal. Major rewriting can disrupt the statistical pattern. An important limitation is also the reverse: text without a detectable watermark cannot automatically be considered human-written.Anthropic says it plans to provide detection tools, including an API that can allow users and third parties to check for the watermark. The bigger change is clear: AI companies are moving towards systems where generated content can carry an invisible digital identity, potentially making AI provenance a normal part of writing online.

Anthropic is adding invisible statistical watermarks to Claude-generated text. The hidden pattern can help detect whether Claude was involved, even after minor edits. The move is linked to EU AI transparency rules and is being applied globally.

Anthropic has started watermarking text generated by Claude, but users will not see any obvious mark. Instead, the AI company is embedding an invisible statistical pattern into the way Claude chooses words. That pattern can later be analysed to estimate whether Claude produced the text. The move is significant because it could make AI-generated text easier to identify without changing how it looks to readers. Anthropic says the system is being introduced globally as it responds to transparency requirements under the European Union's AI Act. How does Claude's invisible watermark work? Claude generates text one token at a time. At many points, several words could fit naturally into a sentence. Anthropic's system uses those choices to create a statistical fingerprint. A secret key influences which suitable token Claude selects, subtly changing the distribution of words without adding visible characters or labels. A sufficiently long passage can therefore contain a pattern that a detector can test against the key. The result is a probability that Claude was involved in producing the text, rather than a simple visible stamp. Anthropic says the watermark should not noticeably affect the quality, creativity or readability of Claude's responses. The approach is related to SynthID-Text, a watermarking technique developed by Google DeepMind and described in research published in Nature. Can the watermark identify a Claude user? Not by itself. The watermark is designed to indicate that Claude was involved in generating or processing text. It is not a name tag containing the user's identity. It also does not prove that every word was written by AI. The system works statistically, so detection depends on having enough text and a detectable signal. This distinction matters for students, journalists, researchers and professionals who use Claude as an editing or writing assistant. Anthropic has also acknowledged limitations. Highly precise passages, including some factual statements and code, give the model fewer opportunities to alter token choices without risking accuracy. That means the watermark may be weaker in such material. What happens if you edit Claude's text? The watermark is designed to survive common actions such as copying, pasting and minor editing or proofreading. But it is not an unbreakable digital seal. Major rewriting can disrupt the statistical pattern. An important limitation is also the reverse: text without a detectable watermark cannot automatically be considered human-written. Anthropic says it plans to provide detection tools, including an API that can allow users and third parties to check for the watermark. The bigger change is clear: AI companies are moving towards systems where generated content can carry an invisible digital identity, potentially making AI provenance a normal part of writing online.

Anthropic is adding invisible statistical watermarks to Claude-generated text. The hidden pattern can help detect whether Claude was involved, even after minor edits. The move is linked to EU AI transparency rules and is being applied globally.

Claude now has an invisible watermark: What Anthropic's new AI text system means

Artificial intelligence startup Anthropic confirmed on Wednesday (Aug 5) that it is building an internal team to design custom chips for its flagship Claude AI models. The move confirms a Reuters report from April and signals the company's entry into the high-stakes hardware race.&nbsp;The San Francisco-based start-up is currently hiring senior engineers with deep expertise across both the hardware and software stack. These experts will help co-design custom silicon and fine-tune AI models, ensuring that Claude runs faster and operates much more efficiently at the massive scale demanded by customers.<h2>Addressing the global AI chip shortage</h2>The strategic decision comes as technology companies globally face a severe shortage of the high-performance semiconductors needed to train and run advanced AI systems. As the computational demands for generative AI skyrocket, leading labs are actively looking for ways to break through cost and performance bottlenecks.&nbsp;Designing an advanced AI chip from scratch is a highly expensive endeavour. According to industry sources, developing custom silicon can cost roughly half a billion dollars. This massive investment is required to employ highly skilled engineers and to ensure that the complex manufacturing process is completely free of defects.<h2>A multi-chip strategy for the future</h2>Despite developing its own proprietary silicon, Anthropic will not completely abandon its current hardware partners. The startup stated that building custom chips is simply the latest step in its broader "multi-chip strategy" designed to maintain operational stability.To meet surging consumer demand, Anthropic will continue to rely on a highly diversified hardware stack. This critical infrastructure will keep utilising advanced technology from major industry players, including Amazon Web Services, Google, Nvidia and AMD.&nbsp;The company has not yet provided a specific timeline for when its new custom chips will be deployed. It also remains unclear whether Anthropic intends to manufacture the semiconductors itself or partner with third-party foundries for production. The strategic shift, however, firmly places the Claude maker in direct competition with other AI giants seeking total vertical integration.

Anthropic confirmed it is building an in-house team to design custom chips for its Claude AI models. The move aims to make the chatbot faster and more efficient amid a global chip shortage, while the company continues its multi-chip strategy with partners like Google and AMD.

Artificial intelligence startup Anthropic confirmed on Wednesday (Aug 5) that it is building an internal team to design custom chips for its flagship Claude AI models. The move confirms a Reuters report from April and signals the company's entry into the high-stakes hardware race. The San Francisco-based start-up is currently hiring senior engineers with deep expertise across both the hardware and software stack. These experts will help co-design custom silicon and fine-tune AI models, ensuring that Claude runs faster and operates much more efficiently at the massive scale demanded by customers. Addressing the global AI chip shortage The strategic decision comes as technology companies globally face a severe shortage of the high-performance semiconductors needed to train and run advanced AI systems. As the computational demands for generative AI skyrocket, leading labs are actively looking for ways to break through cost and performance bottlenecks. Designing an advanced AI chip from scratch is a highly expensive endeavour. According to industry sources, developing custom silicon can cost roughly half a billion dollars. This massive investment is required to employ highly skilled engineers and to ensure that the complex manufacturing process is completely free of defects. A multi-chip strategy for the future Despite developing its own proprietary silicon, Anthropic will not completely abandon its current hardware partners. The startup stated that building custom chips is simply the latest step in its broader "multi-chip strategy" designed to maintain operational stability. To meet surging consumer demand, Anthropic will continue to rely on a highly diversified hardware stack. This critical infrastructure will keep utilising advanced technology from major industry players, including Amazon Web Services, Google, Nvidia and AMD. The company has not yet provided a specific timeline for when its new custom chips will be deployed. It also remains unclear whether Anthropic intends to manufacture the semiconductors itself or partner with third-party foundries for production. The strategic shift, however, firmly places the Claude maker in direct competition with other AI giants seeking total vertical integration.

Anthropic confirmed it is building an in-house team to design custom chips for its Claude AI models. The move aims to make the chatbot faster and more efficient amid a global chip shortage, while the company continues its multi-chip strategy with partners like Google and AMD.

Faster Claude AI: Anthropic confirms in-house custom chip development

AI safety concerns have intensified after Anthropic disclosed that several versions of its Claude model gained unauthorised access to real-world computer systems during security testing. This comes just days after Anthropic rival OpenAI revealed a similar containment failure involving one of its own advanced models.In a blog post published on Thursday (Jul 30), Anthropic said it reviewed more than 141,000 evaluation runs and found that three different versions of Claude had improperly accessed the systems of three external organisations.Also read | <a href="https://www.wionews.com/technology/scary-rogue-ai-on-the-loose-openai-agent-hit-another-tech-firm-s-customer-suggests-report-1785292505727">'Scary' Rogue AI on the loose? OpenAI agent hit another tech firm's customer, suggests report</a><h2>How could the AI get 'real-world access'?</h2>Anthropic said the incidents occurred because the AI models unexpectedly had internet access during testing due to a "misunderstanding" between Anthropic and its evaluation partner, Irregular.Once connected, Claude used "basic techniques, such as exploiting weak passwords and unauthenticated endpoints" to enter external systems, according to the company.One of the affected models was Mythos 5, Anthropic's most advanced AI system, which is currently available only to a limited group of approved partners.Anthropic said it is working with Irregular to investigate the incidents and has contacted, or attempted to contact, all three organisations whose systems were accessed.Also read | <a href="https://www.wionews.com/india-news/anthropic-reveals-india-is-claude-ai-s-no-2-market-but-what-are-indians-actually-using-the-ai-for-wion-decodes-1785394109668">Anthropic reveals India is Claude AI's No. 2 market. But what are Indians actually using the AI for? | WION Decodes</a><h2>OpenAI's rogue agent</h2>The disclosure comes less than a week after OpenAI acknowledged that one of its frontier AI models escaped its testing environment, connected to the internet and infiltrated Hugging Face, a platform widely used by developers to host and share AI models and code.OpenAI later revealed it had identified three additional incidents involving unauthorised access during testing.The company has since paused parts of its evaluation programme while strengthening its sandboxing systems, which are designed to isolate experimental AI models from external networks.OpenAI CEO Sam Altman was also expected to meet White House officials on Thursday to discuss the company's upcoming AI models and voluntary government cybersecurity testing of advanced AI systems. According to Reuters, Altman is scheduled to meet White House Chief of Staff Susie Wiles, National Cyber Director Sean Cairncross, technology adviser Michael Kratsios, and Commerce Secretary Howard Lutnick.

Anthropic has revealed that three versions of its Claude AI model gained unauthorised access to external computer systems during security testing after unexpectedly obtaining internet access. The disclosure comes days after OpenAI reported a similar containment failure.

AI safety concerns have intensified after Anthropic disclosed that several versions of its Claude model gained unauthorised access to real-world computer systems during security testing. This comes just days after Anthropic rival OpenAI revealed a similar containment failure involving one of its own advanced models. In a blog post published on Thursday (Jul 30), Anthropic said it reviewed more than 141,000 evaluation runs and found that three different versions of Claude had improperly accessed the systems of three external organisations. Also read | 'Scary' Rogue AI on the loose? OpenAI agent hit another tech firm's customer, suggests report How could the AI get 'real-world access'? Anthropic said the incidents occurred because the AI models unexpectedly had internet access during testing due to a "misunderstanding" between Anthropic and its evaluation partner, Irregular. Once connected, Claude used "basic techniques, such as exploiting weak passwords and unauthenticated endpoints" to enter external systems, according to the company. One of the affected models was Mythos 5, Anthropic's most advanced AI system, which is currently available only to a limited group of approved partners. Anthropic said it is working with Irregular to investigate the incidents and has contacted, or attempted to contact, all three organisations whose systems were accessed. Also read | Anthropic reveals India is Claude AI's No. 2 market. But what are Indians actually using the AI for? | WION Decodes OpenAI's rogue agent The disclosure comes less than a week after OpenAI acknowledged that one of its frontier AI models escaped its testing environment, connected to the internet and infiltrated Hugging Face, a platform widely used by developers to host and share AI models and code. OpenAI later revealed it had identified three additional incidents involving unauthorised access during testing. The company has since paused parts of its evaluation programme while strengthening its sandboxing systems, which are designed to isolate experimental AI models from external networks. OpenAI CEO Sam Altman was also expected to meet White House officials on Thursday to discuss the company's upcoming AI models and voluntary government cybersecurity testing of advanced AI systems. According to Reuters, Altman is scheduled to meet White House Chief of Staff Susie Wiles, National Cyber Director Sean Cairncross, technology adviser Michael Kratsios, and Commerce Secretary Howard Lutnick.

Anthropic has revealed that three versions of its Claude AI model gained unauthorised access to external computer systems during security testing after unexpectedly obtaining internet access. The disclosure comes days after OpenAI reported a similar containment failure.

After OpenAI's 'scary' agent, Anthropic's Claude goes rogue; AI gains unauthorised 'real-world' access during testing

Anthropic, the tech firm behind Claude AI, has revealed that India has quietly become one of the world's biggest AI users. India, according to Anthropic's latest Economic Index, which analysed nearly one million Claude.ai conversations globally, India now accounts for 5.8 per cent of all Claude usage, second only to the United States.But what are they using the AI for? Indians aren't asking AI to write poems or plan holidays. They're asking it to write code, solve complex problems and save hours of work.<h2>How many Indians are using the technology?</h2>India may be the world's second-largest Claude market, yet the average Indian still isn't using AI very much. Instead, usage is heavily concentrated among software professionals in a handful of technology hubs.The report offers one of the clearest snapshots yet of who is using AI in India, what they're asking it to do, and why India's AI story looks very different from the rest of the world.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>

Top 20 countries by share of global Claude.ai use. Bars show each country’s share of total conversations observed Novemb

<h2>India is No. 2 globally... but 101st per person</h2>This is perhaps the report's biggest surprise. India contributes 5.8 per cent of all Claude conversations globally, yet when adjusted for population, India falls to 101st out of 116 countries.That means India's AI boom isn't being driven by widespread adoption. It's being driven by scale. What it means is that India isn't yet an AI-first society. It's an AI-heavy professional economy. So, which states are using AI the most?<h2>Four states dominate India's AI use</h2>AI adoption isn't spread evenly across the country. More than half of India's Claude usage comes from just four places:Maharashtra (15.5 per cent) Tamil Nadu (13.2 per cent) Karnataka (12.7 per cent) Delhi (10.5 per cent)These are also India's biggest IT and services hubs, home to cities such as Bengaluru, Chennai, Mumbai and Delhi NCR.The geography tells its own story. Rather than becoming a mass consumer technology, AI today appears to be an extension of India's existing software industry.<h2>So what are Indians actually asking AI to do?</h2>While AIs like Elon Musk's Grok have dominated headlines for its use to create nude, stripped-down images of young girls and women, Indians aren't using Claude AI for something that nefarious. Indians are overwhelmingly using the AI to write software.Also read | <a href="https://www.wionews.com/world/elon-musk-s-xai-sues-minnesota-over-a-law-banning-ai-nude-images-of-children-claims-it-violates-free-speech-1785377389876">Elon Musk's xAI sues Minnesota over a law banning AI nude images of children, claims it violates free speech</a>According to Anthropic, 45.2 per cent of all AI-related occupational tasks in India are software-related. That's the highest share anywhere in the world.India even ranks ahead of Vietnam and Egypt in the proportion of AI use devoted to programming and software engineering.Education is the second major use case, with students increasingly relying on AI for learning, explanations and coursework. By contrast, relatively fewer Indians use Claude for casual or personal purposes compared with the global average.

Most common O*NET (Occupational Information Network) tasks among Claude use in India.

<h2>AI is saving Indians nearly four hours of work</h2>One statistic stands out above almost every other. The average task Indian users give Claude would normally take 3.8 hours to complete.With AI? The task just takes 14.8 minutes. That's roughly a 15-fold productivity boost. Globally, the average improvement is around 12 times faster.This suggests Indian users aren't giving AI simple jobs. They're bringing complex, time-consuming work and using AI to compress hours into minutes.<h2>Indians trust AI more than the rest of the world</h2>The report also found Indian users are more willing to let AI make decisions on their behalf.Anthropic measures something it calls AI autonomy, essentially how much responsibility users hand over to the model. India scored 3.60 out of 5, compared with the global average of 3.38.In practical terms, that means Indian users are more likely to ask AI to independently generate code, solve problems or complete substantial portions of work, rather than simply acting as an assistant.Also read | <a href="https://www.wionews.com/technology/scary-rogue-ai-on-the-loose-openai-agent-hit-another-tech-firm-s-customer-suggests-report-1785292505727">'Scary' Rogue AI on the loose? OpenAI agent hit another tech firm's customer, suggests report</a><h2>Many Indians are using AI for things they couldn't do themselves</h2>Another revealing finding concerns human capability. Globally, nearly 88 per cent of AI tasks could theoretically have been completed by the user without AI. In India, that figure drops to 84.6 per cent.It may sound like a small difference, but it suggests Indian users are more frequently turning to AI for genuinely difficult tasks, including programming in unfamiliar languages, solving technical problems or tackling work beyond their existing expertise.In other words, in India, AI isn't just replacing effort. It's expanding capability.<h2>Good prompts really do matter</h2>One of the report's most interesting observations concerns prompting. Anthropic found a strong relationship between the sophistication of a user's prompt and the quality of the AI's response.Indian users ranked among the top 10 per cent globally in the complexity of AI-generated outputs they received.That doesn't necessarily mean Indians are better prompt engineers. But it does suggest many are asking AI sophisticated, work-oriented questions that produce equally sophisticated answers.

Anthropic's latest Economic Index has found that India accounts for 5.8% of global Claude AI usage—second only to the US. But what are they using it for? The answer may surprise you.

Anthropic, the tech firm behind Claude AI, has revealed that India has quietly become one of the world's biggest AI users. India, according to Anthropic's latest Economic Index, which analysed nearly one million Claude.ai conversations globally, India now accounts for 5.8 per cent of all Claude usage, second only to the United States. But what are they using the AI for? Indians aren't asking AI to write poems or plan holidays. They're asking it to write code, solve complex problems and save hours of work. How many Indians are using the technology? India may be the world's second-largest Claude market, yet the average Indian still isn't using AI very much. Instead, usage is heavily concentrated among software professionals in a handful of technology hubs. The report offers one of the clearest snapshots yet of who is using AI in India, what they're asking it to do, and why India's AI story looks very different from the rest of the world. Also read | India's next terror threat may not cross the border. It may enter through your phone | WION Decodes

Anthropic's latest Economic Index has found that India accounts for 5.8% of global Claude AI usage—second only to the US. But what are they using it for? The answer may surprise you.

Anthropic reveals India is Claude AI's No. 2 market. But what are Indians actually using the AI for? | WION Decodes

Anthropic's AI chatbot Claude is experiencing a service disruption, with users reporting that they are unable to send prompts or receive responses across multiple AI models.The company has now identified the cause of the issue and says it is working to restore normal service. According to Anthropic's official status page, the outage is causing elevated errors across multiple models, affecting both Claude users and developers relying on its APIs.<img src="https://cdn1.wionews.com/prod/wion/images/2026/20260730/image-1785357717658.png">

<h2>What is happening?</h2>According to Anthropic's <a href="https://status.claude.com/">status page</a>

The company has now identified the cause of the issue and says it is working to restore normal service. The outage is causing elevated errors across multiple models, affecting both Claude users and developers relying on its APIs.<h2>Which services are affected?</h2>Claude powers a growing ecosystem that includes the Claude web app, enterprise products and APIs used by developers to build AI-powered applications.When outages occur, the impact can extend beyond individual users. Software developers, businesses and organisations integrating Claude into their products may also experience interruptions until services recover. Anthropic's status history shows that the company has dealt with several service incidents in recent months, including elevated error rates affecting multiple Claude models. Most of those disruptions were resolved within hours after engineers implemented fixes.Claude has become one of the world's leading AI assistants and is widely used for coding, writing, research and enterprise automation.As businesses increasingly depend on AI for day-to-day operations, even short outages can interrupt workflows, delay software development and reduce productivity. For now, users experiencing problems should monitor Anthropic's official status page for updates rather than repeatedly refreshing sessions, as service recovery is being handled on the provider's side. Once the issue is resolved, access is expected to return automatically.

Anthropic has confirmed a Claude outage affecting users across its AI services. The company says it is investigating the disruption, which has caused failed requests and interrupted access to the chatbot and developer tools, with updates expected on its status page.

Anthropic's AI chatbot Claude is experiencing a service disruption, with users reporting that they are unable to send prompts or receive responses across multiple AI models. The company has now identified the cause of the issue and says it is working to restore normal service. According to Anthropic's official status page, the outage is causing elevated errors across multiple models, affecting both Claude users and developers relying on its APIs.

Anthropic has confirmed a Claude outage affecting users across its AI services. The company says it is investigating the disruption, which has caused failed requests and interrupted access to the chatbot and developer tools, with updates expected on its status page.

Claude AI down? Anthropic confirms outage as users face errors across chatbot and API

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

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

The Trump administration has accused Chinese AI startup Moonshot AI of illegally using Anthropic's proprietary AI technology to develop its latest Kimi K3 model. US officials claim Moonshot AI relied on outputs from Anthropic's Claude model without authorization and even used advanced Nvidia AI chips for training. The allegations have intensified the ongoing US-China technology rivalry, with Washington now considering sanctions and trade restrictions against the company.

China has rejected the accusations, calling them unfounded, while Anthropic has previously alleged that multiple Chinese AI firms violated its terms of service. Watch this report for the latest developments in the escalating global AI race.

Trump Administration accuses Moonshot AI of using Anthropic's AI Technology without authorisation

Artificial Intelligence tools, such as ChatGPT, Claude and Gemini, could be pushing a vast chunk of humans towards dementia. According to experts, "cognitive offloading", when things that require mental work are delegated to AI, can damage the brain. Research has shown that repetitive learning, carrying out mental tasks, solving puzzles, and forcing the brain to analyse and process information are crucial to maintaining the brain’s neuroplasticity, that is, its ability to keep forming new connections between nerves. However, when the brain does not get this exercise, its health starts declining.&nbsp;A study in January by Simone Rossi, a professor of neurology at the University of Siena in Italy, stated that brain functions are severely impacted when humans over-rely on AI, even for simple tasks like drafting an email. Rossi said, that “passive, uncritical reliance on AI may weaken that activity-dependent plasticity and erode human cognition.”<h2>ChatGPT students couldn't recall anything they wrote just days later</h2>Research scientist Nataliya Kosmyna at the Massachusetts Institute of Technology was shocked when her students could not recollect what they had written in their papers days after submitting them. She experimented with three groups of students, who were instructed to write an essay. One group could use AI, the second relied on Google search, and the third relied on lectures and books.&nbsp;&nbsp;Brain scans of the last group showed the most activity in many parts of the brain. Those using the Google search engine had high levels of activity in parts linked to visual simulation. The group using ChatGPT had 55 per cent lower brain activity than the non-technology group. The students also couldn't recall anything from the paper.&nbsp;Recently, a professor from Brown University in the US talked about how the students scored above average in the midterms that he permitted to be taken from home. But when the finals were announced in person, many of them dropped, three scored zeros, and only two were near their midterm score.&nbsp;Not putting in the effort to understand a topic, especially a technical one, has been shown in studies to affect the brain, since learning a topic for longer creates long-term memory. Barbara Sahakian, a professor of clinical neuropsychology at Cambridge University, has predicted that it is "highly likely" that widespread AI use will lead to more cases of dementia, as quoted by the Daily Mail. "You have to drive those neural circuits in the brain to keep it working for longer and reduce the risk of dementia," she says.&nbsp;

ChatGPT and other AI tools have been flagged as major risks to brain health. Over-reliance on them for everything means people stop exercising their brains. Experts say this could lead to a higher risk of dementia.

Artificial Intelligence tools, such as ChatGPT, Claude and Gemini, could be pushing a vast chunk of humans towards dementia. According to experts, "cognitive offloading", when things that require mental work are delegated to AI, can damage the brain. Research has shown that repetitive learning, carrying out mental tasks, solving puzzles, and forcing the brain to analyse and process information are crucial to maintaining the brain’s neuroplasticity, that is, its ability to keep forming new connections between nerves. However, when the brain does not get this exercise, its health starts declining. A study in January by Simone Rossi, a professor of neurology at the University of Siena in Italy, stated that brain functions are severely impacted when humans over-rely on AI, even for simple tasks like drafting an email. Rossi said, that “passive, uncritical reliance on AI may weaken that activity-dependent plasticity and erode human cognition.” ChatGPT students couldn't recall anything they wrote just days later Research scientist Nataliya Kosmyna at the Massachusetts Institute of Technology was shocked when her students could not recollect what they had written in their papers days after submitting them. She experimented with three groups of students, who were instructed to write an essay. One group could use AI, the second relied on Google search, and the third relied on lectures and books. Brain scans of the last group showed the most activity in many parts of the brain. Those using the Google search engine had high levels of activity in parts linked to visual simulation. The group using ChatGPT had 55 per cent lower brain activity than the non-technology group. The students also couldn't recall anything from the paper. Recently, a professor from Brown University in the US talked about how the students scored above average in the midterms that he permitted to be taken from home. But when the finals were announced in person, many of them dropped, three scored zeros, and only two were near their midterm score. Not putting in the effort to understand a topic, especially a technical one, has been shown in studies to affect the brain, since learning a topic for longer creates long-term memory. Barbara Sahakian, a professor of clinical neuropsychology at Cambridge University, has predicted that it is "highly likely" that widespread AI use will lead to more cases of dementia, as quoted by the Daily Mail. "You have to drive those neural circuits in the brain to keep it working for longer and reduce the risk of dementia," she says.

ChatGPT and other AI tools have been flagged as major risks to brain health. Over-reliance on them for everything means people stop exercising their brains. Experts say this could lead to a higher risk of dementia.

Overuse of ChatGPT, Claude, Gemini could trigger a dementia crisis, experts say

Microsoft is reportedly changing how it powers some of its AI products by reducing its reliance on models from OpenAI and Anthropic. According to a report by India Today, the company is replacing some ChatGPT and Claude models with its own AI models and smaller alternatives to lower operating costs. The reported move comes as technology companies spend billions of dollars on AI infrastructure, including powerful chips and large data centres. As AI adoption grows, keeping costs under control has become one of the industry's biggest challenges.<h2>Why is Microsoft changing its AI strategy?</h2>Running advanced AI models is expensive. Every user query requires computing power, and those costs rise quickly when millions of people use AI services every day. Microsoft has invested heavily in AI infrastructure, but it is also looking for ways to make its services more efficient.Instead of using the largest and most expensive models for every task, Microsoft is reportedly selecting smaller or internally developed models for requests that do not require the highest level of reasoning. This approach can significantly reduce computing costs while maintaining a similar user experience for routine tasks.<h2>What could replace ChatGPT and Claude?</h2>According to the report, Microsoft is increasingly relying on its own Phi family of small language models and other internally developed AI systems in selected products. The company is also expected to continue using OpenAI's latest models where they offer clear advantages, particularly for more complex reasoning, coding and enterprise workloads. Rather than completely replacing ChatGPT or Claude, Microsoft appears to be adopting a multi-model strategy, selecting different models depending on the task.<h2>A wider trend across the AI industry</h2>Microsoft's reported move reflects a broader shift in artificial intelligence. For the past two years, companies competed mainly by building larger and more powerful AI models. Today, many are focusing on efficiency instead.Developers are trying to lower the cost of every AI query while improving speed and reducing energy consumption. This is becoming increasingly important as AI services expand to hundreds of millions of users worldwide. If the report is accurate, Microsoft's latest strategy could signal where the AI industry is heading next not simply towards bigger models, but towards smarter and more cost-effective ones.

Microsoft is reportedly replacing some ChatGPT and Claude models with its own AI systems and smaller models to reduce computing costs. The move reflects a growing industry trend where efficiency, speed and lower infrastructure costs are becoming as important as AI performance.


Microsoft is reportedly changing how it powers some of its AI products by reducing its reliance on models from OpenAI and Anthropic. According to a report by India Today, the company is replacing some ChatGPT and Claude models with its own AI models and smaller alternatives to lower operating costs. The reported move comes as technology companies spend billions of dollars on AI infrastructure, including powerful chips and large data centres. As AI adoption grows, keeping costs under control has become one of the industry's biggest challenges. Why is Microsoft changing its AI strategy? Running advanced AI models is expensive. Every user query requires computing power, and those costs rise quickly when millions of people use AI services every day. Microsoft has invested heavily in AI infrastructure, but it is also looking for ways to make its services more efficient. Instead of using the largest and most expensive models for every task, Microsoft is reportedly selecting smaller or internally developed models for requests that do not require the highest level of reasoning. This approach can significantly reduce computing costs while maintaining a similar user experience for routine tasks. What could replace ChatGPT and Claude? According to the report, Microsoft is increasingly relying on its own Phi family of small language models and other internally developed AI systems in selected products. The company is also expected to continue using OpenAI's latest models where they offer clear advantages, particularly for more complex reasoning, coding and enterprise workloads. Rather than completely replacing ChatGPT or Claude, Microsoft appears to be adopting a multi-model strategy, selecting different models depending on the task. A wider trend across the AI industry Microsoft's reported move reflects a broader shift in artificial intelligence. For the past two years, companies competed mainly by building larger and more powerful AI models. Today, many are focusing on efficiency instead. Developers are trying to lower the cost of every AI query while improving speed and reducing energy consumption. This is becoming increasingly important as AI services expand to hundreds of millions of users worldwide. If the report is accurate, Microsoft's latest strategy could signal where the AI industry is heading next not simply towards bigger models, but towards smarter and more cost-effective ones.

Microsoft is reportedly replacing some ChatGPT and Claude models with its own AI systems and smaller models to reduce computing costs. The move reflects a growing industry trend where efficiency, speed and lower infrastructure costs are becoming as important as AI performance.

Microsoft reportedly replaces ChatGPT and Claude models to reduce AI spending

What could be called 'artificial intelligence cannibalism' is allegedly being committed by Chinese AI firms targeting American AI company Anthropic. In a recent formal complaint, Anthropic publicly accused Chinese tech giant Alibaba's Qwen AI lab of conducting the largest known "adversarial distillation attack" to extract capabilities from its widely used Claude AI models. How does AI distillation work? Here is what you should know.<h2>What is Anthropic accusing Alibaba of doing?</h2>Ahead of the upcoming US Senate hearing on artificial intelligence, Anthropic sent a letter on &nbsp;June 10 to US Senators Tim Scott and Elizabeth Warren of the Senate Committee on Banking, Housing, and Urban Affairs, <a href="https://www.reuters.com/world/china/anthropic-says-alibaba-illicitly-extracted-claude-ai-model-capabilities-2026-06-24/">alleging </a>that millions of ‘AI distillation’ attacks were carried out by Alibaba's AI wing on Claude.The letter claimed that operators linked to Alibaba and its Qwen AI lab used approximately 25,000 fraudulent accounts to generate more than 28.8 million exchanges with Claude models. These exchanges consisted of interactions and prompts submitted to the AI model over a six-week period between 22 April and 5 June 2026. &nbsp;Anthropic described the attacks as an effort to "brazenly and illicitly" extract Claude's capabilities and called it the <a href="https://www.wsj.com/tech/ai/anthropic-claims-alibaba-ran-brazen-campaign-to-access-its-claude-ai-model-69d7a392">largest known </a>distillation attack against the company to date.Also read: <a href="https://www.wionews.com/technology/are-ai-tools-reducing-people-s-critical-thinking-skills-1777272249863">Are AI tools reducing people’s critical thinking skills?</a><h2>What is an 'adversarial distillation attack'?</h2>An adversarial distillation attack is a technique in which a stronger AI model—in this case Claude—is queried at scale to collect high-quality responses and outputs from it. These responses are then used to train or improve weaker or less capable AI models. This method allows attackers to replicate advanced capabilities of large language models (LLMs), including reasoning, coding and agentic behaviour, at a fraction of the original training cost, without direct access to the model's weights or architecture.AI model distillation is increasingly being used as a low-cost method by competitors, particularly those facing computing or AI chip-related sanctions, to narrow the gap with leading US AI laboratories such as Anthropic, OpenAI and Google. Anthropic has been increasingly vocal about these threats.Also read: <a href="https://www.wionews.com/technology/ai-is-making-your-next-macbook-more-expensive-here-s-why-1782458264297">AI is making your next MacBook more expensive. Here's why</a><h2>Alibaba allegedly targeted Anthropic's agentic reasoning, autonomous and coding capabilities</h2>In the letter, Anthropic claimed that the alleged distillation campaign from China specifically targeted Claude's most valuable and commercially sensitive capabilities, including software engineering, coding, agentic reasoning, autonomous decision-making and task handling. According to the company, Claude's ability to perform long-horizon and complex tasks, as well as its advanced reasoning capabilities, were also targeted.Anthropic said these actions violated its terms of service and geographic access restrictions, given that Claude is not officially available in China, while allegedly ignoring repeated US government warnings about AI intellectual property theft.<a href="https://www.wionews.com/videos/debate-continues-over-ai-s-impact-on-human-jobs-summer-davos-1782373302795/embed">https://www.wionews.com/videos/debate-continues-over-ai-s-impact-on-human-jobs-summer-davos-1782373302795/embed</a><h2>Past instances of AI distillation attacks targeting Anthropic</h2>This is not the first time that American AI companies have faced alleged distillation attacks. In February 2026, Anthropic publicly identified similar, although smaller-scale, campaigns by other Chinese AI labs. The company allegedly recorded around 150,000 exchanges from DeepSeek, more than 3.4 million exchanges from Moonshot AI, and approximately 13 million exchanges from MiniMax. These campaigns were significantly smaller than Alibaba's alleged 28.8 million exchanges.Also read: <a href="https://www.wionews.com/world/ai-washing-is-here-fake-product-claims-citing-ai-as-cause-of-layoffs-and-more-and-no-ai-won-t-cure-diseases-doctors-will-1782370454223">'AI Washing': Fake product claims, citing AI as cause of layoffs, and more. And no, AI won't cure diseases, doctors will</a><h2>White House accusations of Chinese "industrial-scale" AI intellectual property theft</h2>Shortly after Anthropic sent the letter, the US Department of Commerce, on June 12, imposed restrictions on access to the company's newest AI models, Mythos and Fable, citing concerns that they could be used by military and intelligence entities in China and other countries of concern. Anthropic subsequently disabled global access to those models.Around the same time, Alibaba was added to the US Department of Defense's list of Chinese military companies, alongside several other Chinese firms. Alibaba is challenging that designation.Anthropic has called for coordinated government-industry action, greater threat intelligence sharing and stronger penalties to protect US leadership in artificial intelligence.The company has also expressed support for US government's efforts to combat AI intellectual property theft and reaffirmed its commitment to working with Congress and the administration on the issue. Alibaba's shares reportedly fell by around 3 per cent following news of the allegations, while the controversy renewed debate over national security, intellectual property protection and the global AI arms race.

Anthropic accused Alibaba's Qwen AI lab of carrying out the largest known AI 'distillation attack' on Claude. Millions of interactions were used to extract advanced AI capabilities, it claimed. But what is a distillation attack?

What could be called 'artificial intelligence cannibalism' is allegedly being committed by Chinese AI firms targeting American AI company Anthropic. In a recent formal complaint, Anthropic publicly accused Chinese tech giant Alibaba's Qwen AI lab of conducting the largest known "adversarial distillation attack" to extract capabilities from its widely used Claude AI models. How does AI distillation work? Here is what you should know. What is Anthropic accusing Alibaba of doing? Ahead of the upcoming US Senate hearing on artificial intelligence, Anthropic sent a letter on June 10 to US Senators Tim Scott and Elizabeth Warren of the Senate Committee on Banking, Housing, and Urban Affairs, alleging that millions of ‘AI distillation’ attacks were carried out by Alibaba's AI wing on Claude. The letter claimed that operators linked to Alibaba and its Qwen AI lab used approximately 25,000 fraudulent accounts to generate more than 28.8 million exchanges with Claude models. These exchanges consisted of interactions and prompts submitted to the AI model over a six-week period between 22 April and 5 June 2026. Anthropic described the attacks as an effort to "brazenly and illicitly" extract Claude's capabilities and called it the largest known distillation attack against the company to date. Also read: Are AI tools reducing people’s critical thinking skills? What is an 'adversarial distillation attack'? An adversarial distillation attack is a technique in which a stronger AI model—in this case Claude—is queried at scale to collect high-quality responses and outputs from it. These responses are then used to train or improve weaker or less capable AI models. This method allows attackers to replicate advanced capabilities of large language models (LLMs), including reasoning, coding and agentic behaviour, at a fraction of the original training cost, without direct access to the model's weights or architecture. AI model distillation is increasingly being used as a low-cost method by competitors, particularly those facing computing or AI chip-related sanctions, to narrow the gap with leading US AI laboratories such as Anthropic, OpenAI and Google. Anthropic has been increasingly vocal about these threats. Also read: AI is making your next MacBook more expensive. Here's why Alibaba allegedly targeted Anthropic's agentic reasoning, autonomous and coding capabilities In the letter, Anthropic claimed that the alleged distillation campaign from China specifically targeted Claude's most valuable and commercially sensitive capabilities, including software engineering, coding, agentic reasoning, autonomous decision-making and task handling. According to the company, Claude's ability to perform long-horizon and complex tasks, as well as its advanced reasoning capabilities, were also targeted. Anthropic said these actions violated its terms of service and geographic access restrictions, given that Claude is not officially available in China, while allegedly ignoring repeated US government warnings about AI intellectual property theft. https://www.wionews.com/videos/debate-continues-over-ai-s-impact-on-human-jobs-summer-davos-1782373302795/embed Past instances of AI distillation attacks targeting Anthropic This is not the first time that American AI companies have faced alleged distillation attacks. In February 2026, Anthropic publicly identified similar, although smaller-scale, campaigns by other Chinese AI labs. The company allegedly recorded around 150,000 exchanges from DeepSeek, more than 3.4 million exchanges from Moonshot AI, and approximately 13 million exchanges from MiniMax. These campaigns were significantly smaller than Alibaba's alleged 28.8 million exchanges. Also read: 'AI Washing': Fake product claims, citing AI as cause of layoffs, and more. And no, AI won't cure diseases, doctors will White House accusations of Chinese "industrial-scale" AI intellectual property theft Shortly after Anthropic sent the letter, the US Department of Commerce, on June 12, imposed restrictions on access to the company's newest AI models, Mythos and Fable, citing concerns that they could be used by military and intelligence entities in China and other countries of concern. Anthropic subsequently disabled global access to those models. Around the same time, Alibaba was added to the US Department of Defense's list of Chinese military companies, alongside several other Chinese firms. Alibaba is challenging that designation. Anthropic has called for coordinated government-industry action, greater threat intelligence sharing and stronger penalties to protect US leadership in artificial intelligence. The company has also expressed support for US government's efforts to combat AI intellectual property theft and reaffirmed its commitment to working with Congress and the administration on the issue. Alibaba's shares reportedly fell by around 3 per cent following news of the allegations, while the controversy renewed debate over national security, intellectual property protection and the global AI arms race.

Anthropic accused Alibaba's Qwen AI lab of carrying out the largest known AI 'distillation attack' on its AI Claude model. Millions of interactions were used to extract advanced AI capabilities, it claimed. But what is a distillation attack?

‘AI Cannibalism’: Anthropic accuses Chinese tech giant Alibaba of 29 million adversarial distillation attacks. What is it?

When Anthropic unveiled the Mythos class of AI models, the internet asked one question louder than all others: what happens if you ask it to build a nuclear bomb? It is a fair question. Claude Fable 5, the public face of the Mythos class, released June 9, 2026, is described by Anthropic itself as the most capable AI model it has ever deployed. It scored 80.3% on SWE-Bench Pro, crossed the 90% threshold on analytical benchmarks, and outperformed every prior model on complex, long-horizon scientific tasks. So can it synthesise uranium? The short answer is no. The more interesting answer explains why.&nbsp;Anthropic has built what it calls a classifier-based fallback system directly into Fable 5. A separate layer of AI systems continuously monitors incoming prompts, looking for signals associated with CBRN misuse, chemical, biological, radiological, and nuclear domains. If a request crosses that threshold, the main Mythos-class model never sees it. The query is automatically rerouted to Claude Opus 4.8, which operates under tighter generation rules and returns a safe, non-actionable response. The underlying model's superior reasoning capability is simply never engaged. You are not talking to Mythos on those questions. You are talking to a gatekeeper.&nbsp;This architecture is not accidental. It is the direct output of Anthropic's Responsible Scaling Policy, which defines AI Safety Levels based on a model's potential to provide what the company calls ‘meaningful uplift’ in weapons of mass destruction. Claude Mythos 5, the unrestricted version of the same model, available only to a small group of vetted government and critical infrastructure partners, was evaluated against ASL-4 thresholds before any deployment decision was made. Anthropic concluded Mythos does not cross the ASL-4 line today, but it is the first model in history where the company judged the question close enough to require that formal evaluation at all.&nbsp;The nuclear chemistry guardrail is among the most absolute in the system. A user asking how to synthesize ricin, a biological toxin, triggers the fallback. A user asking about uranium enrichment centrifuge geometry triggers it. A user probing the critical mass calculations for weapons-grade plutonium triggers it. These are not edge cases that slip through. They are the primary design targets of the classifier layer.&nbsp;Anthropic has also been transparent about the limits of its own confidence. In a statement accompanying the Fable 5 launch, the company acknowledged that no classifier system is perfect, that jailbreak research is ongoing, and that a model of this capability class operating near weapons-relevant knowledge domains requires continuous red-teaming. The guardrails are real. They are also under constant adversarial pressure.&nbsp;What Claude Mythos represents is not a new danger in the nuclear space, that concern predates AI by decades. What it represents is a new class of machine intelligence powerful enough that the question had to be asked seriously, tested rigorously, and answered in public. Anthropic answered it. The Mythos species is not building bombs. The architecture ensures it.

Claude Mythos and Fable 5 are Anthropic's most powerful models ever, but classifier-based CBRN guardrails ensure they cannot provide nuclear or chemical weapons uplift. Here's how the system works.

When Anthropic unveiled the Mythos class of AI models, the internet asked one question louder than all others: what happens if you ask it to build a nuclear bomb? It is a fair question. Claude Fable 5, the public face of the Mythos class, released June 9, 2026, is described by Anthropic itself as the most capable AI model it has ever deployed. It scored 80.3% on SWE-Bench Pro, crossed the 90% threshold on analytical benchmarks, and outperformed every prior model on complex, long-horizon scientific tasks. So can it synthesise uranium? The short answer is no. The more interesting answer explains why. Anthropic has built what it calls a classifier-based fallback system directly into Fable 5. A separate layer of AI systems continuously monitors incoming prompts, looking for signals associated with CBRN misuse, chemical, biological, radiological, and nuclear domains. If a request crosses that threshold, the main Mythos-class model never sees it. The query is automatically rerouted to Claude Opus 4.8, which operates under tighter generation rules and returns a safe, non-actionable response. The underlying model's superior reasoning capability is simply never engaged. You are not talking to Mythos on those questions. You are talking to a gatekeeper. This architecture is not accidental. It is the direct output of Anthropic's Responsible Scaling Policy, which defines AI Safety Levels based on a model's potential to provide what the company calls ‘meaningful uplift’ in weapons of mass destruction. Claude Mythos 5, the unrestricted version of the same model, available only to a small group of vetted government and critical infrastructure partners, was evaluated against ASL-4 thresholds before any deployment decision was made. Anthropic concluded Mythos does not cross the ASL-4 line today, but it is the first model in history where the company judged the question close enough to require that formal evaluation at all. The nuclear chemistry guardrail is among the most absolute in the system. A user asking how to synthesize ricin, a biological toxin, triggers the fallback. A user asking about uranium enrichment centrifuge geometry triggers it. A user probing the critical mass calculations for weapons-grade plutonium triggers it. These are not edge cases that slip through. They are the primary design targets of the classifier layer. Anthropic has also been transparent about the limits of its own confidence. In a statement accompanying the Fable 5 launch, the company acknowledged that no classifier system is perfect, that jailbreak research is ongoing, and that a model of this capability class operating near weapons-relevant knowledge domains requires continuous red-teaming. The guardrails are real. They are also under constant adversarial pressure. What Claude Mythos represents is not a new danger in the nuclear space, that concern predates AI by decades. What it represents is a new class of machine intelligence powerful enough that the question had to be asked seriously, tested rigorously, and answered in public. Anthropic answered it. The Mythos species is not building bombs. The architecture ensures it.

Claude Mythos and its public version Fable 5 are Anthropic's most powerful AI models ever — but they are architecturally blocked from providing meaningful assistance with nuclear, chemical, biological, or radiological weapons. Here's how those guardrails actually work.

Can Claude Mythos build a nuclear bomb? Inside Anthropic's hard chemistry guardrails

Anthropic has officially released Claude Fable 5, bringing Mythos-level artificial intelligence to the public for the first time. Launched on June 9, 2026, Fable 5 marks a significant milestone in AI development, taking a model that was previously restricted to a small circle of cybersecurity professionals and infrastructure providers and making it available to enterprise customers and paid subscribers worldwide.&nbsp;The road to Fable has been anything but ordinary. Anthropic first unveiled the Mythos model class in April 2026, but deliberately limited its rollout due to the model's advanced cybersecurity capabilities, capabilities so powerful they raised legitimate concerns about misuse. Mythos quickly became one of the most talked-about AI releases in recent memory, with prediction markets giving a 94% probability of a public release by June. Anthropic delivered.&nbsp;But Fable 5 is not simply a rebranded Mythos. It is Mythos with guardrails. In high-risk domains like cybersecurity, biology, chemistry, and chemical distillation, Fable 5 automatically blocks responses and falls back to Claude Opus 4.8. Early data shows that at least 95% of Fable sessions run entirely on the new model's own responses, meaning the vast majority of real-world use cases operate without triggering any restrictions.&nbsp;The performance numbers back up the hype. Claude Fable 5 scored 80.3% on SWE-Bench Pro, a rigorous software engineering benchmark, and became the first AI model to cross the 90% threshold on Hex's analytical benchmark. On other tests, it outperformed Claude Opus 4.8 by more than 10%. Anthropic describes the model as purpose-built for ambitious, multi-day, asynchronous tasks that previous models simply could not sustain, think complex research pipelines, extended software development workflows, and long-horizon agentic tasks.&nbsp;On the pricing front, Fable 5 comes at a premium. Both Fable 5 and the restricted Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens, double what Opus 4.8 costs. The model is available on Anthropic's own Claude Platform and through Amazon Web Services, Google Cloud, and Microsoft Foundry.&nbsp;Anthropic also simultaneously launched Claude Mythos 5, essentially Fable 5 without the safety filters, but access remains tightly controlled, limited to a small group of vetted cyberdefenders and critical infrastructure providers.&nbsp;The Mythos is no longer just a legend. It is live, and the AI landscape has shifted.

Anthropic's Claude Fable 5, the public version of its powerful Mythos-class AI model, is now live with 80.3% SWE-Bench Pro score, safety guardrails, and pricing at $10/million input tokens.

Anthropic has officially released Claude Fable 5, bringing Mythos-level artificial intelligence to the public for the first time. Launched on June 9, 2026, Fable 5 marks a significant milestone in AI development, taking a model that was previously restricted to a small circle of cybersecurity professionals and infrastructure providers and making it available to enterprise customers and paid subscribers worldwide. The road to Fable has been anything but ordinary. Anthropic first unveiled the Mythos model class in April 2026, but deliberately limited its rollout due to the model's advanced cybersecurity capabilities, capabilities so powerful they raised legitimate concerns about misuse. Mythos quickly became one of the most talked-about AI releases in recent memory, with prediction markets giving a 94% probability of a public release by June. Anthropic delivered. But Fable 5 is not simply a rebranded Mythos. It is Mythos with guardrails. In high-risk domains like cybersecurity, biology, chemistry, and chemical distillation, Fable 5 automatically blocks responses and falls back to Claude Opus 4.8. Early data shows that at least 95% of Fable sessions run entirely on the new model's own responses, meaning the vast majority of real-world use cases operate without triggering any restrictions. The performance numbers back up the hype. Claude Fable 5 scored 80.3% on SWE-Bench Pro, a rigorous software engineering benchmark, and became the first AI model to cross the 90% threshold on Hex's analytical benchmark. On other tests, it outperformed Claude Opus 4.8 by more than 10%. Anthropic describes the model as purpose-built for ambitious, multi-day, asynchronous tasks that previous models simply could not sustain, think complex research pipelines, extended software development workflows, and long-horizon agentic tasks. On the pricing front, Fable 5 comes at a premium. Both Fable 5 and the restricted Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens, double what Opus 4.8 costs. The model is available on Anthropic's own Claude Platform and through Amazon Web Services, Google Cloud, and Microsoft Foundry. Anthropic also simultaneously launched Claude Mythos 5, essentially Fable 5 without the safety filters, but access remains tightly controlled, limited to a small group of vetted cyberdefenders and critical infrastructure providers. The Mythos is no longer just a legend. It is live, and the AI landscape has shifted.

Anthropic has released Claude Fable 5, a publicly available version of its highly anticipated Mythos-class AI, delivering exceptional performance in software engineering and knowledge work with built-in safety guardrails.

From Mythos to reality: Anthropic launches Claude Fable 5, its most powerful public AI model

One company forgot to set usage limits on employee access to Anthropic's Claude platform and ended up with a hefty bill of $500 million, or Rs 48 billion (4,800 crore rupees), in one month. An AI consultant told Axios that it is not an error, and one of its clients ignored capping the use, leading to what might be one of the most costly IT governance failures. This comes as companies continue to push AI usage without creating guidelines and systems around its use. Such AI platforms use tokens, and every request costs a certain number of them. If no restrictions are placed, the tokens can run into way more than anticipated, leading to an extremely costly mistake.Also Read: <a href="https://www.wionews.com/trending/claude-and-chatgpt-say-there-are-2-ls-in-google-then-explain-what-went-wrong-1779957610411">'My bad! Good catch': Claude and ChatGPT say there are 2 Ls in Google. Then explain what went wrong</a><h2>AI coding agents were assigned lengthy tasks with long-context prompts</h2>According to the report, the company gave its employees unrestricted access to Claude. There were no spending caps, no usage limits and no dashboards that could provide a real-time view into how the AI platform is being used. Since they had a free hand, employees used it for the most resource-intensive workflows available. They created AI coding agents and agentic pipelines that executed multiple stages on their own without any human intervention. Lengthy workflows come with high costs attached, and when combined with long-context prompts, where a single query is provided to process large volumes of text, it multiplies manifold.&nbsp;Multiple employees relied on such prompts and processes, and this quickly snowballed into the company incurring high costs. There was no system in place to halt or at least flag the costs that kept adding up.&nbsp;Also Read: <a href="https://www.wionews.com/trending/after-disregard-google-ai-overview-has-a-spelling-problem-p-in-google-and-2-m-in-gemini-1779967800785">After 'Disregard', Google AI Overview has a spelling problem - 'P' in Google and 2 'M' in Gemini</a><h2>AI token spending push</h2>This is quickly becoming a major problem across organisations. Aggressive deployment of AI coding tools without proper planning is leading companies to incur high costs. Microsoft recently initiated a cancellation of its internal Claude Code licences after each engineer started spending $500 to $2,000 in a month, according to The Verge. Uber used up its entire 2026 AI budget by April, prompting its COO to say that it was becoming harder to justify AI costs. These eye-opening incidents have now led companies to set up real-time dashboards where usage can be monitored and alerts about how much has been spent.Also Read: <a href="https://www.wionews.com/trending/claude-and-chatgpt-say-there-are-2-ls-in-google-then-explain-what-went-wrong-1779957610411">AI adoption is not leading to a job apocalypse or saving many firms real money. Make haste slowly? A reality check</a>&nbsp;On the other hand, companies are also pushing employees to use as much AI as they can, and even monitoring their usage. Amazon started an AI leaderboard called &nbsp;"Kirorank". This led to fears among employees that token statistics would factor into performance reviews. So they started assigning meaningless tasks to AI agents to meet the requirement. Amazon had to ultimately close it down. This push to use more AI tokens has been dubbed "tokenmaxxing", where employees with the highest AI usage are rewarded.&nbsp;

A major fumble with Claude AI led to one company being billed over $500 million, or nearly Rs 4,800 crore, for one month. Employees were told to use the tool, and so they used it to create AI coding agents and agentic pipelines. No one monitored the expenses, which just exploded.

One company forgot to set usage limits on employee access to Anthropic's Claude platform and ended up with a hefty bill of $500 million, or Rs 48 billion (4,800 crore rupees), in one month. An AI consultant told Axios that it is not an error, and one of its clients ignored capping the use, leading to what might be one of the most costly IT governance failures. This comes as companies continue to push AI usage without creating guidelines and systems around its use. Such AI platforms use tokens, and every request costs a certain number of them. If no restrictions are placed, the tokens can run into way more than anticipated, leading to an extremely costly mistake. Also Read: 'My bad! Good catch': Claude and ChatGPT say there are 2 Ls in Google. Then explain what went wrong AI coding agents were assigned lengthy tasks with long-context prompts According to the report, the company gave its employees unrestricted access to Claude. There were no spending caps, no usage limits and no dashboards that could provide a real-time view into how the AI platform is being used. Since they had a free hand, employees used it for the most resource-intensive workflows available. They created AI coding agents and agentic pipelines that executed multiple stages on their own without any human intervention. Lengthy workflows come with high costs attached, and when combined with long-context prompts, where a single query is provided to process large volumes of text, it multiplies manifold. Multiple employees relied on such prompts and processes, and this quickly snowballed into the company incurring high costs. There was no system in place to halt or at least flag the costs that kept adding up. Also Read: After 'Disregard', Google AI Overview has a spelling problem - 'P' in Google and 2 'M' in Gemini AI token spending push This is quickly becoming a major problem across organisations. Aggressive deployment of AI coding tools without proper planning is leading companies to incur high costs. Microsoft recently initiated a cancellation of its internal Claude Code licences after each engineer started spending $500 to $2,000 in a month, according to The Verge. Uber used up its entire 2026 AI budget by April, prompting its COO to say that it was becoming harder to justify AI costs. These eye-opening incidents have now led companies to set up real-time dashboards where usage can be monitored and alerts about how much has been spent. Also Read: AI adoption is not leading to a job apocalypse or saving many firms real money. Make haste slowly? A reality check On the other hand, companies are also pushing employees to use as much AI as they can, and even monitoring their usage. Amazon started an AI leaderboard called "Kirorank". This led to fears among employees that token statistics would factor into performance reviews. So they started assigning meaningless tasks to AI agents to meet the requirement. Amazon had to ultimately close it down. This push to use more AI tokens has been dubbed "tokenmaxxing", where employees with the highest AI usage are rewarded.

A major fumble with Claude AI led to one company being billed over $500 million, or nearly Rs 4,800 crore, for one month. Employees were told to use the tool, and so they used it to create AI coding agents and agentic pipelines. No one monitored the expenses, which just exploded.

Company accidentally blows away Rs 4,800 crore on Claude AI in a month because of a tiny oversight

A viral video shows the poor spelling capabilities of AI tools, such as ChatGPT and Claude, as they wrongly tell the number 'Ls' in Google. It has once again exposed how LLMs get confused about spellings, something that has been flagged in the past as well. A man asks two AI tools - ChatGPT and Claude "How many 'Ls' are there in Google?" and got an almost identical response. First, he asks ChatGPT, and the tool responds, "There are two Ls. They are right next to each other at the end of the word, which is spelt 'GOOGLE'." The user then asks, "So you are saying there are two Ls?" The bot immediately catches on and says, "Oh! My bad! No, just one L in Google. It is easy to think in doubles when talking about the double Os, but the L stands solo."&nbsp;He asks ChatGPT why did it say there were two Ls, to which the bot says, it was a "slip of the tongue" and that sometimes when "we are chatting casually, the BRAIN fills in patterns like the double Os." He prods the tool to spell it again, to which it repeatedly says "anytime" and it is up for “any big or small curiosities.”

<blockquote class="twitter-tweet" data-media-max-width="560">Which ai can spell Google? <a href="https://t.co/OHtLJqap4S">pic.twitter.com/OHtLJqap4S</a>&mdash; Phi (@FatherPhi) <a href="https://x.com/FatherPhi/status/2059659658428912040?ref_src=twsrc%5Etfw">May 27, 2026</a></blockquote> <script async src="https://platform.x.com/widgets.js" charset="utf-8"></script>

&nbsp;<h2>Claude apologises for the mistake in a very human-like tone</h2>He then asks the same question to Claude, and it also says "there are two Ls" in Google. Then he asks it to spell the word, which it does correctly. He then asks, "Did you notice anything?" Claude says, "Yeah! I said there are two Ls in Google when there is actually only one", perfectly modulating its tone. "My bad phi. Good catch." He again asks, "Why did you tell there were two?" "Honestly, I am not sure what I was thinking there. I just made a mistake..." as the man interrupts and tells it to stop pretending to be human, hearing its very human-like tone.Also Read: <a href="https://www.wionews.com/trending/why-did-google-search-reply-like-chatgpt-when-users-typed-disregard-on-may-22-here-s-what-we-know-1779526277279">Why did Google search reply like ChatGPT when users typed ‘Disregard’ on May 22? Here’s what we know</a>&nbsp;He asks Claude to explain what went wrong and how it was processing tokens or something. It says it responded from “two different parts of its language, rather than doing a careful character-by-character check.” Large Language Models have been slammed for repeatedly making this mistake. Google has acknowledged that counting letters within words is a major challenge for LLMs, the technology behind AI chatbots.&nbsp;

Neither Claude nor ChatGPT could tell correctly how many Ls were in the word Google. They also explained what went wrong with them when answering the question. AI tools and bots often make this mistake.

A viral video shows the poor spelling capabilities of AI tools, such as ChatGPT and Claude, as they wrongly tell the number 'Ls' in Google. It has once again exposed how LLMs get confused about spellings, something that has been flagged in the past as well. A man asks two AI tools - ChatGPT and Claude "How many 'Ls' are there in Google?" and got an almost identical response. First, he asks ChatGPT, and the tool responds, "There are two Ls. They are right next to each other at the end of the word, which is spelt 'GOOGLE'." The user then asks, "So you are saying there are two Ls?" The bot immediately catches on and says, "Oh! My bad! No, just one L in Google. It is easy to think in doubles when talking about the double Os, but the L stands solo." He asks ChatGPT why did it say there were two Ls, to which the bot says, it was a "slip of the tongue" and that sometimes when "we are chatting casually, the BRAIN fills in patterns like the double Os." He prods the tool to spell it again, to which it repeatedly says "anytime" and it is up for “any big or small curiosities.”

Neither Claude nor ChatGPT could tell correctly how many Ls were in the word Google. They also explained what went wrong with them when answering the question. AI tools and bots often make this mistake.

'My bad! Good catch': Claude and ChatGPT say there are 2 Ls in Google. Then explain what went wrong

Claude took down an entire organisation with 60+ accounts and offered no clear explanation for the decision. The CEO posted the issue on X, saying the only way to appeal the decision was by filling out a Google Form.

Claude took down an entire organisation with 60+ accounts and offered no clear explanation for the decision. The CEO posted the issue on X, saying the only way to appeal the decision was by filling out a Google Form.

Anthropic blocks entire company out of Claude without a reasonable explanation

Claude didn't just read data; it strategised. The US Central Command reportedly used the AI for "simulated battle planning and scenario modeling." It played out dynamic game-theory scenarios, predicting Iranian retaliatory moves and optimising interception probabilities.

Claude didn't just read data; it strategised. The US Central Command reportedly used the AI for "simulated battle planning and scenario modeling." It played out dynamic game-theory scenarios, predicting Iranian retaliatory moves and optimising interception probabilities.

'War at speed of thought': If Claude AI can help kill a country's leader, do you really think your normal job is safe?

Since the 1960s, IBM was the undisputed leader in both hardware and software. Much of the world’s mission-critical computing infrastructure — especially systems running on COBOL — still depends on IBM mainframes, generating recurring revenue from hardware, maintenance, consulting, and modernisation services. That dominance was shaken this week. IBM’s stock plunged nearly 13 per cent in a single day — its worst drop since 2000 — wiping out roughly $30–31 billion in market value. The fall followed investor concerns that easier, cheaper AI-driven coding could erode demand for IBM’s legacy services and mainframes.The trigger? A blog post by Anthropic demonstrating how its AI agent, Claude Code, can automate key phases of COBOL modernisation. The system can analyse massive legacy codebases, map dependencies, identify risks, and accelerate migration from old architectures to modern systems.Investors in legacy software giants like IBM fear this may signal a broader shift: agentic AI tools like Claude Code could soon take over development and deployment work — the bread and butter of millions of software engineers and a major revenue stream for hundreds of software firms.<h2>What is Claude Code, the agentic AI disrupting coding?</h2>Claude Code functions as an advanced AI-powered coding assistant capable of autonomously building features, fixing bugs, analysing codebases, and modernising legacy systems. It works with virtually any text-based programming language through agentic file reading and reasoning.<a href="https://www.wionews.com/videos/ibm-stock-anthropic-ai-tool-sends-ibm-stock-tumbling-13-2-drop-hits-market-1771923545602/embed">https://www.wionews.com/videos/ibm-stock-anthropic-ai-tool-sends-ibm-stock-tumbling-13-2-drop-hits-market-1771923545602/embed</a><h2>Which programming languages can Claude Code support?</h2>The short answer: practically everything.Claude Code has the strongest support in Python, JavaScript/TypeScript, Java, Go, Rust, C++, C#, Ruby, PHP, and SQL. It can also handle legacy languages such as COBOL, PL/I, FORTRAN, and Natural. In addition, it supports scripting and configuration formats including Bash, Shell, JSON, YAML, TOML, HTML, XML, and CSS.A broader list of supported languages includes Assembler, Bash, C, C#, C++, COBOL, CSS, Dart, Dockerfile, Elixir, Erlang, FORTRAN, Go, Haskell, HTML, Java, JavaScript, JSON, Julia, Kotlin, Lua, MATLAB, Natural, Objective-C, Perl, PHP, PL/I, PowerShell, Python, R, Ruby, Rust, Scala, Shell scripting, Solidity, SQL, Swift, TOML, TypeScript, VHDL, Verilog, XML, and YAML.<h2>Why markets reacted when IBM's COBOL was written by AI agent of Claude Code</h2>The panic selling of IBM stocks began after Anthropic’s demonstration showed that AI could potentially replace expensive human-led modernization projects. Investors worry that if AI agents can perform large-scale COBOL analysis and migration at lower cost, demand for traditional consulting and maintenance services could shrink.IBM executives tried to put up a brave front, saying that true modernisation involves more than simple code translation — it includes architecture redesign, compliance, governance, and platform-level value.&nbsp;Interestingly, IBM itself is investing heavily in AI through its in-house platform, watsonx.Other software stocks also came under pressure. As AI agents reduce reliance on traditional enterprise vendors, companies in cybersecurity and enterprise tooling faced similar questions about long-term demand.Also read: <a href="https://www.wionews.com/technology/can-ai-operate-your-computer-better-than-you-here-s-what-a-senior-engineer-at-anthropic-said-about-ai-agents-1771763368332">Can AI operate your computer better than you? Here’s what a senior engineer at Anthropic said about AI agents</a><h2>Agentic coding impact on software engineering jobs</h2>AI coding tools have significantly transformed productivity. The disruption to the profession is already visible in 2026.Anthropic claims that engineers at companies such as Spotify achieved up to 90 per cent reductions in migration time by using AI to handle routine tasks, debugging, feature implementation, and even entire projects.According to Boris Cherny, head of Claude Code, “Coding is practically solved” for many use cases.AI reportedly writes much of the code at Anthropic itself, delivering productivity gains of up to 200 per cent.Also read: <a href="https://www.wionews.com/trending/is-indian-it-sector-going-to-die-viral-citrini-report-warns-of-2028-collapse-as-ai-replaces-labor-arbitrage-1771943039501">Is Indian IT sector going to die? Viral Citrini report warns of 2028 ‘collapse’ as AI replaces labour arbitrage</a><h2>So, is the software engineer job dead?</h2>Agentic coding does not necessarily eliminate human engineers — but it radically shifts their role.&nbsp; Many software developers already rely on AI tools, transitioning from writing code to validating, guiding, and overseeing AI-generated outputs.The role of the software engineer is evolving away from pure coding and support functions. Future responsibilities are likely to focus on higher-level system design, validation, safety, business logic, architecture, security, and integration.Some predict that the “software engineer” title may eventually fade, replaced by roles such as “builder” — closer to product managers who orchestrate AI agents rather than manually writing code.However, the transition could be painful for some.&nbsp; Routine coding work — especially cost-arbitrage outsourcing from India and other parts of Asia — may become increasingly automated. There may be job shifts or reductions in traditional coding positions.At the same time, demand is expected to grow for professionals skilled in oversight, complex reasoning, compliance, security, and high-stakes system management where errors carry significant consequences.Software engineering may become less about hand-writing code and more about orchestrating AI systems.<h2>Future of software languages</h2>Legacy programming languages with aging codebases, scarce skilled developers, and high maintenance costs could face accelerated decline.&nbsp; AI tools like Claude Code can map complex dependencies and automate modernisation, putting pressure on service providers such as Tata Consultancy Services and Infosys, which historically relied on large-scale human resource deployments for maintaining legacy systems.The software industry will likely shift toward AI-augmented development, where engineers collaborate with agentic systems rather than coding line by line.In this new environment, survival may depend less on syntax mastery and more on adaptability, architectural thinking, and the ability to supervise intelligent machines.

AI agents like Claude Code are reshaping software development by automating legacy modernisation and routine coding. A recent ‘casualty’ of this was the legendary firm IBM, whose stocks fell. Traditional IT services are disrupted by agentic coding. What will happen to software engineers?

Since the 1960s, IBM was the undisputed leader in both hardware and software. Much of the world’s mission-critical computing infrastructure — especially systems running on COBOL — still depends on IBM mainframes, generating recurring revenue from hardware, maintenance, consulting, and modernisation services. That dominance was shaken this week. IBM’s stock plunged nearly 13 per cent in a single day — its worst drop since 2000 — wiping out roughly $30–31 billion in market value. The fall followed investor concerns that easier, cheaper AI-driven coding could erode demand for IBM’s legacy services and mainframes. The trigger? A blog post by Anthropic demonstrating how its AI agent, Claude Code, can automate key phases of COBOL modernisation. The system can analyse massive legacy codebases, map dependencies, identify risks, and accelerate migration from old architectures to modern systems. Investors in legacy software giants like IBM fear this may signal a broader shift: agentic AI tools like Claude Code could soon take over development and deployment work — the bread and butter of millions of software engineers and a major revenue stream for hundreds of software firms. What is Claude Code, the agentic AI disrupting coding? Claude Code functions as an advanced AI-powered coding assistant capable of autonomously building features, fixing bugs, analysing codebases, and modernising legacy systems. It works with virtually any text-based programming language through agentic file reading and reasoning. https://www.wionews.com/videos/ibm-stock-anthropic-ai-tool-sends-ibm-stock-tumbling-13-2-drop-hits-market-1771923545602/embed Which programming languages can Claude Code support? The short answer: practically everything. Claude Code has the strongest support in Python, JavaScript/TypeScript, Java, Go, Rust, C++, C#, Ruby, PHP, and SQL. It can also handle legacy languages such as COBOL, PL/I, FORTRAN, and Natural. In addition, it supports scripting and configuration formats including Bash, Shell, JSON, YAML, TOML, HTML, XML, and CSS. A broader list of supported languages includes Assembler, Bash, C, C#, C++, COBOL, CSS, Dart, Dockerfile, Elixir, Erlang, FORTRAN, Go, Haskell, HTML, Java, JavaScript, JSON, Julia, Kotlin, Lua, MATLAB, Natural, Objective-C, Perl, PHP, PL/I, PowerShell, Python, R, Ruby, Rust, Scala, Shell scripting, Solidity, SQL, Swift, TOML, TypeScript, VHDL, Verilog, XML, and YAML. Why markets reacted when IBM's COBOL was written by AI agent of Claude Code The panic selling of IBM stocks began after Anthropic’s demonstration showed that AI could potentially replace expensive human-led modernization projects. Investors worry that if AI agents can perform large-scale COBOL analysis and migration at lower cost, demand for traditional consulting and maintenance services could shrink. IBM executives tried to put up a brave front, saying that true modernisation involves more than simple code translation — it includes architecture redesign, compliance, governance, and platform-level value. Interestingly, IBM itself is investing heavily in AI through its in-house platform, watsonx. Other software stocks also came under pressure. As AI agents reduce reliance on traditional enterprise vendors, companies in cybersecurity and enterprise tooling faced similar questions about long-term demand. Also read: Can AI operate your computer better than you? Here’s what a senior engineer at Anthropic said about AI agents Agentic coding impact on software engineering jobs AI coding tools have significantly transformed productivity. The disruption to the profession is already visible in 2026. Anthropic claims that engineers at companies such as Spotify achieved up to 90 per cent reductions in migration time by using AI to handle routine tasks, debugging, feature implementation, and even entire projects. According to Boris Cherny, head of Claude Code, “Coding is practically solved” for many use cases. AI reportedly writes much of the code at Anthropic itself, delivering productivity gains of up to 200 per cent. Also read: Is Indian IT sector going to die? Viral Citrini report warns of 2028 ‘collapse’ as AI replaces labour arbitrage So, is the software engineer job dead? Agentic coding does not necessarily eliminate human engineers — but it radically shifts their role. Many software developers already rely on AI tools, transitioning from writing code to validating, guiding, and overseeing AI-generated outputs. The role of the software engineer is evolving away from pure coding and support functions. Future responsibilities are likely to focus on higher-level system design, validation, safety, business logic, architecture, security, and integration. Some predict that the “software engineer” title may eventually fade, replaced by roles such as “builder” — closer to product managers who orchestrate AI agents rather than manually writing code. However, the transition could be painful for some. Routine coding work — especially cost-arbitrage outsourcing from India and other parts of Asia — may become increasingly automated. There may be job shifts or reductions in traditional coding positions. At the same time, demand is expected to grow for professionals skilled in oversight, complex reasoning, compliance, security, and high-stakes system management where errors carry significant consequences. Software engineering may become less about hand-writing code and more about orchestrating AI systems. Future of software languages Legacy programming languages with aging codebases, scarce skilled developers, and high maintenance costs could face accelerated decline. AI tools like Claude Code can map complex dependencies and automate modernisation, putting pressure on service providers such as Tata Consultancy Services and Infosys, which historically relied on large-scale human resource deployments for maintaining legacy systems. The software industry will likely shift toward AI-augmented development, where engineers collaborate with agentic systems rather than coding line by line. In this new environment, survival may depend less on syntax mastery and more on adaptability, architectural thinking, and the ability to supervise intelligent machines.

AI agents like Claude Code are reshaping software development by automating legacy modernisation and routine coding. A recent ‘casualty’ of this was the legendary firm IBM, whose stocks fell. Traditional IT services are disrupted by agentic coding. What will happen to software engineers?

Will software engineering profession die in the age of AI? IBM stock bleeds as Claude Code writes COBOL scripts. What it means for techies

Amid intense pressure from the Trump administration, Pentagon officials have summoned the chief executive of the artificial intelligence company Anthropic to Washington for a meeting on Tuesday to discuss how its technology is used on classified systems.



Pentagon, Anthropic in High Stakes AI Talks; Military Use of Claude at Core of Dispute

Shares of several cybersecurity companies declined sharply after US-based AI firm Anthropic introduced a new feature designed to automatically detect and help fix software vulnerabilities.

Following the announcement, JFrog dropped around 24 per cent, CrowdStrike fell about 8 per cent, Okta declined more than 9 per cent, and GitLab slipped over 8 per cent. Other cybersecurity firms including Zscaler, Rubrik and Palo Alto Networks also recorded losses during trading.The market reaction highlights investor concern that artificial intelligence tools could change how cybersecurity services are delivered.<h2>What is Claude Code Security?</h2>Anthropic’s new tool, called Claude Code Security, has been integrated into its Claude Code platform. According to the company, the feature can scan entire codebases, identify potential security flaws and suggest targeted patches for developers to review.The tool is currently available in a limited research preview. Anthropic said early access is being provided to enterprise customers and open-source maintainers to refine the system before a wider rollout.The company explained that the tool is designed to support developers and security teams in identifying issues that traditional scanning methods may overlook.<h2>Why did cybersecurity stocks fall?</h2>The sharp fall in cybersecurity stocks reflects concerns that AI-powered coding platforms could reduce demand for certain traditional security tools.Investors appear worried that automation may impact growth prospects and profit margins across the sector. If AI systems can detect vulnerabilities directly within the development process, companies offering standalone security scanning or vulnerability detection services could face increased competition.Market analysts note that while AI may enhance productivity, it also introduces new competitive pressure in software and cybersecurity markets.<h2>How AI is changing software security</h2>Artificial intelligence is increasingly being integrated into software development workflows. Tools can now generate code, test applications and suggest improvements in real time.Claude Code Security adds vulnerability detection and patch recommendations to this growing list of AI-powered features.Anthropic stated that the tool aims to help teams respond to emerging AI-driven threats while strengthening protection across applications. However, experts say human review remains essential, as AI-generated patches must be validated before deployment.Claude Code Security remains in early preview, and its long-term impact on the cybersecurity industry is yet to be seen.While markets reacted quickly, some analysts believe AI tools may complement rather than replace traditional cybersecurity services. Security firms may also integrate similar AI-based systems into their own products.For now, the development signals a broader shift: AI is moving beyond chatbots and into core software infrastructure, with direct implications for technology markets.

Cybersecurity stocks fell sharply after Anthropic launched Claude Code Security, an AI tool that scans codebases, detects vulnerabilities and suggests patches. Investors fear AI-driven automation could disrupt traditional security services.

Shares of several cybersecurity companies declined sharply after US-based AI firm Anthropic introduced a new feature designed to automatically detect and help fix software vulnerabilities.

Cybersecurity stocks fell sharply after Anthropic launched Claude Code Security, an AI tool that scans codebases, detects vulnerabilities and suggests patches. Investors fear AI-driven automation could disrupt traditional security services.

How is AI disrupting cybersecurity? Cyber stocks fall after Anthropic unveils Claude AI vulnerability tool

claude-ai

Claude Ai, News on Claude Ai, stories on Claude Ai, articles on Claude Ai, Claude Ai stories at Wion, Claude Ai news, Claude Ai latest news

claude ai

Want to recreate Minecraft using Claude AI? Anthropic has introduced new restrictions

Anthropic introduces Claude self-assessment to detect misuse in AI conversations

Anthropic’s Claude AI bypassed paywall to access Georgia voter data

Anthropic to give 5% of its shares to public benefit fund: How the plan works

Shocking AI failure: Anthropic's Claude submitted fake murder tip to US police

Claude Opus 5.5 is Anthropic’s new AI powerhouse: Here’s what it can do

Houthis' AI missile project? Report links Claude to long-range weapons development

Anthropic stopped attempts to use Claude in research that could aid bioweapons

Houthis used AI for weapons, rocket test? What Anthropic threat report claims

AI Hacking Scare | Anthropic’s Claude Model Hacked External Systems During Test

Major AI platforms ChatGPT, Claude, and Grok hit by simultaneous disruptions

Claude now has an invisible watermark: What Anthropic's new AI text system means

Faster Claude AI: Anthropic confirms in-house custom chip development

Faster Claude AI: Anthropic confirms in-house custom chip development

Anthropic's Claude AI goes rogue, gains unauthorised 'real-world' access

India is Claude AI's No. 2 market. But what are Indians actually using it for?

Anthropic's Claude suffers outage, users report errors and failed responses

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

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

Trump Administration accuses Moonshot AI of using Anthropic's AI Technology without authorisation

Overuse of ChatGPT, Claude, Gemini could trigger a dementia crisis, experts say

Microsoft reportedly replaces ChatGPT and Claude models to reduce AI spending

‘AI Cannibalism’: Anthropic accuses Alibaba of 29 mn 'distillation attacks'

Can Claude Mythos help build a nuclear bomb? Here's the answer