Newsletters
Anthropic suspended access to its two most advanced artificial intelligence (AI) models, Fable 5 and Mythos 5, on Saturday.
This came after the U.S. Commerce Department issued an export control directive ordering the company to block all foreign nationals from accessing the systems.
The directive that, due to compliance requirements, forced the company to disable the models globally for all users, including American users.
The shutdown, ordered just days after Fable 5's public launch and affecting hundreds of millions of users, marks the most significant government intervention in frontier AI deployment to date.
Anthropic said it received the directive from Commerce Secretary Howard Lutnick at 5:21 p.m. ET Friday.
The order, citing national security authorities, banned all foreign nationals, including foreign national employees of Anthropic itself, from accessing Fable 5 and Mythos 5.
"The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance," the company said in a statement.
Access to all other Anthropic models was not affected.
The government's stated concern was a discovered method of "jailbreaking" Fable 5, bypassing its safety guardrails in a manner that could aid hacking.
Anthropic said the directive did not explain the specific concern in detail, but the company's understanding was that the method involved "asking the model to read a specific codebase and fix any software flaws."
Anthropic rejected the government's assessment as a "misunderstanding" and said the issues shown to it were already known, "minor," and available through other publicly accessible AI models, specifically naming OpenAI's GPT-5.5 as a model with equivalent capability.
It said no tester, including U.S. government testers, the UK's AI safety body, and other groups, had found a "universal jailbreak" of Fable 5 during thousands of hours of pre-release safety testing.
"We suspect that perfect jailbreak resistance is not currently possible for any model provider," Anthropic said.
"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. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers," the company noted.
The company said it believed the government should have the power to block unsafe AI deployments "as part of a statutory process that is transparent, fair, clear and grounded in technical facts. This action does not adhere to those principles."
Mythos 5, the unrestricted, more powerful version of Fable 5, had been available to select companies and government partners since April, with federal agencies experimenting with the model ahead of its wider release.
Several officials had raised concerns about its potential use to exploit cybersecurity vulnerabilities in critical infrastructure, including the global financial system.
Fable 5, released publicly on Tuesday, was a locked-down version of Mythos 5 with significant guardrails designed to prevent dangerous cyber and biology applications.
Trump signed an executive order earlier this year that stopped short of creating a mandatory licensing regime for AI models but established a voluntary review framework.
Anthropic is also engaged in a separate legal dispute with the Pentagon, which designated the company a "supply-chain risk" after a contracting dispute over guardrails Anthropic had insisted on spelling out in contracts.
Trump has publicly called Anthropic's leadership "left-wing nut jobs."
Set as preferred source
Set as preferred source
3 min read
