Anthropic shut down its Fable 5 and Mythos 5 AI models after a US Commerce Department directive cited national security concerns over a reported jailbreak vulnerability.

What happened: Anthropic disabled access to Fable 5 and Mythos 5 models for all customers Friday night, days after launching them. The company received a US Commerce Department directive Friday evening that imposed export controls restricting the models' use outside the United States. Anthropic said abruptly disabling the two models was necessary to ensure immediate compliance with the government order.

Why it matters: The Trump administration reportedly grew concerned after learning of a jailbreak that bypassed the models' safety guardrails, allowing them to respond to prompts about cybersecurity, chemistry, and biology. The administration requested a pause in the release to give time for what it called the 'national security apparatus' to be 'hardened' against this kind of threat. This shows how government oversight can force companies to withdraw new AI products even shortly after launch.

What to watch: An administration official suggested to Axios that the hardening effort could be complete 'in the next few weeks,' implying that could be the timeframe for when such concerns might be resolved. Anthropic said the government provided only 'verbal evidence of a potential narrow, non-universal jailbreak' and noted that the vulnerability involved finding 'minor' and 'relatively simple' software flaws—capabilities the company said other publicly available models like GPT-5.5 also possess.