US Lifts Curbs on Anthropic AI Models

The move allows Claude Fable 5 and Mythos 5 to return after a global access shutdown tied to national security concerns.

MIT SMR Editors Reading Time: 2 minutes 

Topics

  • News

    1. US Lifts Curbs on Anthropic AI Models

    2. Tata Communications Plans $152 Million India-Singapore Subsea Cable Expansion

    3. RBI Flags AI Cyberattacks As Top Financial Stability Risk

    4. Apple Speeds Up Security Updates As AI Threats Rise

    5. Anthropic, OpenAI Push AI Deeper Into Scientific Research

    6. BAT Leans On ITC Infotech Amid Tech-Led Overhaul

    • The Trump administration has lifted export controls on Anthropic’s Claude Fable 5 and Mythos 5 models, allowing the AI company to begin restoring access after a June order forced it to shut the systems down worldwide.

    Anthropic said the US Department of Commerce had removed the controls on 30 June. Fable 5 will be available globally from 1 July across Claude Platform, Claude.ai, Claude Code and Claude Cowork, while Mythos 5 access has been restored for a set of US organizations and is being expanded through Anthropic’s Project Glasswing program.

    The company had suspended both models on 12 June after the US government, citing national security authorities, ordered it to restrict access by foreign nationals, including non-US Anthropic employees.

    Anthropic said at the time that the order gave no specific details beyond national security concerns, but appeared to relate to a possible bypass, or jailbreak, of Fable 5’s safeguards.

    US commerce secretary Howard Lutnick told Anthropic that a license was no longer required for the export, re-export or in-country transfer of Fable 5 and Mythos 5, WIRED reported, citing a letter it reviewed.
    Lutnick said the company had agreed to detect and address model security risks, work with the US government on standards for current and future models, and report malicious activity.

    The frontier model developer had shut down access to Claude Fable 5 and Mythos 5 in June, following orders from the US President Donald Trump administration to restrict foreign nationals, including company employees, from accessing the models.

    Anthropic said it had trained an improved safety classifier to block the behavior cited in the government’s concerns. It said the new system blocked the reported bypass in more than 99% of cases, though it could also reject some benign coding and debugging requests.

    The reversal comes amid growing concern in the US over Chinese progress in AI-enabled cybersecurity. At the ISC.AI 2026 conference in Beijing on 24 June, 360 Security Technology unveiled Yitian Tulong, a pair of AI security tools for vulnerability discovery and automated defense.

    Founder Zhou Hongyi described Tulongfeng as a “Chinese Mythos” and said it had found 3,432 software flaws, including 105 confirmed by the Chinese government. Those claims have not been independently verified.

    Sridhar Vembu, former chief executive and chief scientist of Zoho Corp., said on X that the restoration of Anthropic’s latest models showed Chinese open-source models had become a serious market threat to US frontier model developers.

    “Competition is always important,” he wrote, adding that American AI models facing serious competition from China was important for India and other countries.

    Vembu said there was no reason for India to be despondent, citing work under way in academia and industry and falling model-training costs.

    Topics

    About the Author


    View More

    Tags:

    More Like This

    You must to post a comment.

    First time here? : Comment on articles and get access to many more articles.