The U.S. government orders Anthropic to disable advanced AI models for foreign nationals amid cybersecurity concerns.

WASHINGTON- The U.S. government has taken aggressive action against Anthropic. Officials ordered the company to revoke access to its most advanced AI systems for all foreign nationals.

The U.S. Department of Commerce issued the directive late Friday. It falls under export control rules. The order forces Anthropic to disable its newest models. These include Claude Fable 5 and Mythos 5. “The net effect… is that we must abruptly disable” the systems, the company said in a statement. The rule applies globally. It covers non-U.S. citizens inside and outside the country. It also blocks foreign employees inside Anthropic from using the models.

‘Mythos’ model sparks security fears

The crackdown targets Anthropic’s “Mythos-class” engine. While experts say the system has advanced cybersecurity capabilities, they further warn that it can find and exploit code flaws at high speed. Some describe it as autonomous. Reports say concerns escalated after tests by Amazon researchers. They reportedly bypassed safeguards in Fable 5.

According to The Wall Street Journal, those findings triggered alarm in Washington. Andy Jassy reportedly alerted the White House. He warned that a “jailbreak” could let hostile actors plan cyberattacks on U.S. infrastructure.

Anthropic pushes back on government action

Anthropic disputes the claims. It calls the response excessive. The company says the evidence was limited. It describes the issue as a “narrow, non-universal jailbreak.” Anthropic further adds that the system identified only minor, already known bugs. “We disagree,” the company said. It warned that such strict standards could “halt all new model deployments.” Anthropic argues the policy could slow the entire AI industry.

Tensions with Washington escalate

The order represents a significant shift. It follows a growing conflict between Anthropic and the Trump administration. Earlier this year, the Pentagon labeled the company a “supply chain risk.” Officials blacklisted it. The dispute began after Anthropic refused military use of its AI. That included autonomous weapons and domestic surveillance. In response, Anthropic filed a lawsuit. It called the blacklist “unprecedented and unlawful.”

The crackdown signals a new phase in AI regulation.