U.S. government orders Anthropic to shut down two of its most powerful AI models – safety warnings backfire

The U.S. government on Friday issued an order requiring Anthropic to shut down access to two of its most powerful AI models, Claude Fable 5 and Claude Mythos 5, effective immediately. The directive, which Anthropic said it received at 5:21 pm ET, is based on national security concerns and applies to all users worldwide, not just foreign nationals targeted by export control rules. Access to other Anthropic models remains unaffected.

Mythos is Anthropic's most capable AI model, previewed in early April and kept under tight restrictions due to its exceptional ability to identify security vulnerabilities in software. According to Anthropic, Mythos found flaws in every major operating system and web browser it tested. Instead of a broad release, the company launched a controlled program called Project Glasswing, sharing the model with about 50 vetted organizations, including Amazon, Apple, Google, Microsoft, and CrowdStrike, for defensive cybersecurity purposes.

Fable 5, released just three days ago, is Anthropic's commercial version of Mythos with added guardrails that block responses in high-risk areas like cybersecurity and biology. The company argued it was safe for general release, and benchmark tests from Vals AI confirmed it immediately became the most capable publicly available AI model.

Anthropic, in a blog post, stated that while the government framed the order as an export control action, the underlying concern appears to be a claimed jailbreak of Fable 5. The company says the government provided only verbal evidence of a "potential narrow, non-universal jailbreak" that essentially involves prompting the model to read a specific codebase and identify flaws. Anthropic argues this capability is already widely available in other public models, such as OpenAI's GPT-5.5, and is routinely used by cybersecurity professionals for defensive purposes.

"We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people," Anthropic wrote. "If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers."

The situation is a major irony for Anthropic, which has built its public identity as a safety-conscious alternative to rivals like OpenAI. OpenAI CEO Sam Altman had earlier criticized Anthropic's handling of Mythos as "fear-based marketing," suggesting that warning about the model's dangers would attract government scrutiny. That prediction has now come true, potentially complicating Anthropic's expected IPO later this year.