Anthropic Pulls Fable 5 and Mythos 5 Worldwide After US Export Control Directive

Just three days after launch, Anthropic's most capable AI models have been pulled offline for everyone on the planet, after the United States government ordered the company to block all foreign nationals from accessing them. Anthropic disabled access to its Fable 5 and Mythos 5 models on June 12 following a US government export control directive citing national security concerns, with the order requiring suspension for all users including foreign national employees, due to concerns about a potential jailbreaking method.

The launch itself had been a big moment for the company. Fable 5 was Anthropic's first broadly available Mythos class model, described as the most capable model the company had ever released to the public , with large gains in software engineering, knowledge work, vision, scientific research and long running autonomous tasks, while Mythos 5 was the same underlying model with some safeguards lifted for trusted cybersecurity and biology users. Before launch, the company said the models underwent thousands of hours of testing involving government agencies, third party researchers and internal teams.

Then came the abrupt reversal . Anthropic said the directive arrived at 5:21pm ET on June 12, did not provide specific details of its national security concern, and that the company's understanding is the government believes it has become aware of a method of jailbreaking Fable 5. According to Anthropic, the actual technique amounts to little more than asking the model to scan code for bugs. The potential jailbreak disclosed to Anthropic essentially consists of asking the model to read a specific codebase and fix any software flaws, and Anthropic reviewed a demonstration that identified only a small number of previously known, minor vulnerabilities that other publicly available models can also discover without any bypass.

Anthropic's pushback has been unusually pointed for a company complying with a federal order. The company maintained no researcher has identified a universal jailbreak capable of broadly bypassing the models' safeguards, argued the cybersecurity capabilities cited by officials already exist in competing frontier systems, and called the action a misunderstanding that is not transparent or fair. Anthropic also warned that pulling the model over a narrow vulnerability could set a problematic precedent that would halt AI model deployments across the industry.

What makes the episode messier is the timing and context . Anthropic had just spent the week explaining why Fable needed broad safeguards, visible and invisible, because the model was unusually capable and dangerous, only for the government to apply an even broader restriction on top of that. The order also follows an earlier rupture between Anthropic and the Trump administration over the company refusing to let the military use its models for domestic surveillance and autonomous weapons, with the Pentagon's CIO framing this latest move as national security over " revenue cycles, clickbait, and pre-IPO valuation. "

The international reaction has been swift. The UK's AI minister flagged that the pause hit British users too, framing it as a matter of technological sovereignty , a sign that allied governments are reading this as the US flexing control over frontier AI rather than a narrow fix.