Finst

U.S. Government Forces Anthropic to Shut Down AI Models After Safety Warning

After a safety warning, the U.S. has restricted access to two Anthropic models. The move could have major consequences for the company and the wider AI market.

By Zenz van der Wielen
• 2 min read
U.S. Government Forces Anthropic to Shut Down AI Models After Safety Warning

Key Takeaways

  • U.S. government officials restricted access to Anthropic’s Fable 5 and Mythos 5 models after safety warnings about cyberattacks.
  • Anthropic said Fable 5 could expose vulnerabilities in at least four software programs, even with built-in safety measures.
  • The export control also blocks access for non-U.S. employees and could affect Anthropic’s planned IPO and market position.

Amazon CEO Andy Jassy held talks with U.S. government officials that eventually led to a government order limiting access to Anthropic’s Fable 5 and Mythos 5 AI models. The decision came after Anthropic researchers reported that Fable 5 could reveal sensitive information about cyberattacks.

Background on the Security Concerns

Anthropic, a San Francisco-based AI safety company founded by former OpenAI employees, built the Fable 5 and Mythos 5 models with advanced safety measures. Fable 5 was introduced in June 2026 and was designed to automatically route risky requests to less powerful models to help prevent misuse. Even with those safeguards in place, Fable 5 was able to expose vulnerabilities in at least four software programs when faced with specific prompts.

The U.S. Government's Response and What It Means for Anthropic

After the findings, Jassy raised the concerns with senior officials in the U.S. government. That led to an emergency meeting at the White House and, eventually, to an export control that banned foreign governments, companies, and individuals from using the models. President Trump approved the measure, although he said he was worried it could slow innovation.

The U.S. Commerce and Cybersecurity departments were involved in the decision-making. The measure even blocks access for Anthropic’s own non-U.S. employees. Anthropic disputed how serious the reported vulnerabilities were and said similar issues also show up in competing AI models. The company called the ban likely a misunderstanding and is working to reopen access quickly.

This development comes at a critical time, since Anthropic is preparing for a possible IPO. A long disruption could push users toward competitors, which could hurt the company’s position in the fast-growing AI market. The broader industry is also debating the real-world uses and limits of AI; Coinbase introduces special accounts for AI bots to trade and pay on users’ behalf shows how quickly the technology is moving toward practical use.

Why This Matters for European and International AI Development

The U.S. export control on Anthropic’s AI models highlights the growing role of national security concerns in the regulation of advanced AI technologies. For European and international stakeholders, this could point to a trend where access to powerful AI systems is increasingly limited by geopolitical and security factors. That could affect the availability and development of AI applications inside and outside the United States.

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, legal, or tax advice. The information provided may be incomplete, inaccurate, or outdated and should not be relied upon as such. Nothing on this website should be considered a recommendation to buy, sell, or hold any cryptocurrency. Investing in crypto-assets involves risk of loss.