Anthropic warns U.S. foreign-access rule could freeze frontier model deployments

Anthropic disabled access to its most advanced AI models after the U.S. government ordered the company to block all foreign nationals from using them, including foreign-national employees inside the United States, according to the Los Angeles Times, Newsweek and DW.

The order targeted Anthropic’s Fable 5 and Mythos 5 models. The Los Angeles Times reported that the Commerce Department sent the letter and that Anthropic shut off access to both systems for all customers to ensure compliance.

Newsweek reported that Anthropic described the order as an “export control directive” covering foreign nationals “whether inside or outside the United States, including foreign national Anthropic employees.” DW reported the same foreign-national scope and said Fable 5 and Mythos 5 were released on June 9, with Fable 5 widely available to the public and Mythos 5 mostly restricted to vetted organisations.

The company said the directive appeared to stem from national-security concerns over a possible jailbreak vulnerability in Fable 5, according to Newsweek. Anthropic said it had investigated and found only a handful of previously known weaknesses that other publicly available models could also discover, while acknowledging that watertight jailbreak protection is probably not currently possible for any model provider.

An unfetched Reuters excerpt says Anthropic warned: “If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.” Because the full Reuters article was not fetched, the quote is treated here as reported context rather than independently verified fetched text.

The practical conflict is between a security rule written around nationality and an AI industry built around international staff, customers, infrastructure and release cycles. If model access must be restricted not only by location but by citizenship or immigration status, companies may have to redesign internal permissions, customer access, employee workflows, compliance checks and deployment plans before releasing new systems.

The Los Angeles Times reported that never before had the U.S. government taken such sweeping measures to restrict foreign access to frontier AI models developed by an American company. It noted that earlier U.S. controls had limited access abroad to technologies such as semiconductors and supercomputers, while restrictions on AI software itself raised constitutional and commercial concerns.

Newsweek reported sharp reaction across the tech sector, with some critics calling the ban unworkable and others warning it could weaken U.S. competition with China in AI. DW said the block marked a major escalation in Anthropic’s confrontation with the Trump administration after failed negotiations over possible U.S. military and intelligence use of its technology, and described the order as a potential setback for Anthropic’s planned public listing.

The supplied evidence does not include the government letter, the exact legal authority, the full Commerce Department rationale, or any final court or regulatory response. It also does not show whether the order will remain limited to Anthropic or become a template for other frontier model providers.

What is verified is a new collision between AI security policy and the operating model of frontier labs. A rule aimed at preventing foreign access to powerful systems has already led one company to shut access for everyone, and the industry now faces a concrete question: whether national-security controls can be applied to model access without freezing the release machinery they are meant to govern.