Anthropic’s Fable 5 and Mythos 5 AI Models Suddenly Go Offline — Here is Why

Anthropic has suspended access to its latest AI models Fable 5 and Mythos 5 after receiving a US government export control directive related to national security concerns, according to a company blog post.

The company said the order requires it to block access to foreign nationals, including employees outside the United States, which has resulted in a global suspension of the two models. Access to other Anthropic models has not been affected.

Anthropic said it received the directive at 5:21 pm ET. The government did not provide detailed reasons for the decision, but the company believes it may be linked to concerns over a potential method of bypassing safeguards in Fable 5.

Company Disputes Security Concerns

In its blog post, Anthropic said it reviewed a demonstration of the reported jailbreak technique and found only minor vulnerabilities. It added that similar issues can also be found in other widely available AI models.

The company said, “We have not even received a disclosure of a concerning non-universal potential jailbreak that led to a harmful result,” and added that no tester had identified a universal jailbreak capable of broadly bypassing safeguards.

Anthropic said the government had only provided verbal evidence of a narrow potential jailbreak involving code review and bug fixing. It added that the capability highlighted is already present in other models used in cybersecurity work.

The company said it followed a “defence in depth” approach while developing Fable 5, combining safety filters and monitoring systems to reduce risks while allowing controlled use.

Anthropic said it disagreed with the decision and believes it is based on a misunderstanding. It added that it is working to restore access as soon as possible.

Regulatory Pressure

The move comes amid growing scrutiny of Anthropic in the United States. Earlier this year, the US Department of Defense reportedly classified the company as a supply chain risk, limiting its use in military systems. Anthropic is currently challenging that designation in court.

Separately, the company has confidentially filed for a public listing earlier this month. It was recently valued at around $965 billion in a funding round.