FOLLOW US

On June 13, Commerce Secretary Howard Lutnick sent a letter to Anthropic CEO Dario Amodei directing the company to cut off access to Fable 5 and Mythos 5 for any foreign national, whether inside or outside the United States. The directive cited export control authorities and national security concerns. Anthropic received it at 5:21 p.m Eastern. By that evening, both models were down for every user on the planet.

Fable 5 Was Three Days Old

Fable 5 launched on June 9. It was Anthropic's first public release in its Mythos class of models. Three days later, it was gone.

The Commerce Department's concern centered on a jailbreak. According to an administration official who spoke to Axios, another company demonstrated to the government that it could bypass Fable 5's protections.

Anthropic says the threat was overstated. The jailbreak is narrow and non-universal, the company argues, and the same capabilities are available through competing AI models. "The level of capability displayed there is widely available from other models," Anthropic wrote in its official statement.

The Backfire

Anthropic built its identity on safety. For months, the company described Mythos as extraordinarily capable of identifying vulnerabilities across major operating systems and browsers. Those warnings were meant to justify the caution with which Anthropic approached its release. They also told the government exactly what it might fear.

Sam Altman, CEO of OpenAI, had flagged this dynamic in April. He called Anthropic's approach "fear-based marketing," arguing it amounted to saying "we have a bomb" and then selling protection.

What Anthropic Says

Anthropic is complying with the order. It is also pushing back publicly.

"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" the company wrote.

The argument goes further. Anthropic says it cannot implement nationality restrictions on access fast enough to comply with the directive's terms. Rather than attempt a partial rollout, it took both models down entirely. The company says it is working to restore access as quickly as possible.

On the broader principle, Anthropic warned that applying this standard across the industry "would essentially halt all new model deployments." Perfect jailbreak resistance, it noted, "is not currently possible for any model provider."

The company also pointed out that it had not received specific details from the government about what, exactly, the national security concern was.

Who Is Affected Now

The suspension is specific to Fable 5 and Mythos 5 models, not all Claude models. Researchers, enterprise clients, and developers who accessed those models through Anthropic's API or through Amazon Web Services lost access on Friday. AWS was required to revoke access across all cloud regions under the same directive.

Some of the researchers who built Fable 5 and Mythos 5 are foreign nationals. Under the terms of the directive, they are not permitted to access the models they helped create.

The Bigger Picture

This is not the first sign of tension between Anthropic and the current administration. The Department of Defense previously designated Anthropic as a supply chain risk. Amazon CEO Andy Jassy spoke with U.S. officials about AI-related concerns earlier this year.

The Commerce Department's action sets a new historical precedent: a federal agency has now demonstrated that it can compel a private AI company to take its flagship product offline, globally, within hours.

Anthropic says the concern is overstated and the government says otherwise. And for now, the government has the authority to act while that dispute plays out.

Fable 5 was three days old. Mythos 5 was still in controlled access. Both are gone. The question is whether the standard that took them down will now be applied to every new AI model that follows, Is this the first step toward a new era of AI regulation, or just a clash of business and political interests?