Anthropic abruptly disabled two of its newest artificial-intelligence models Friday after receiving a directive from the U.S. Commerce Department, marking one of the most aggressive government interventions yet in the rapidly escalating battle over advanced AI systems and national security.
The decision affects Fable 5 and Mythos 5, models Anthropic had introduced only days earlier as part of its latest generation of cybersecurity-focused AI technology. According to Axios, the company shut down access after receiving a notice from Commerce Secretary Howard Lutnick directing restrictions on foreign access to the systems.
The move highlights a broader shift within the Trump administration, which has increasingly treated cutting-edge artificial intelligence as a strategic national-security asset rather than a purely commercial technology. The dispute also arrives as Anthropic prepares for a public-market debut following its confidential initial public offering filing earlier this month.
In a statement posted late Friday, Anthropic said the government directive arrived unexpectedly.
"We received the directive from the government today at 5:21pm (ET). The letter did not provide specific details of its national security concern. Our understanding is that the government believes it has become aware of a method of bypassing, or 'jailbreaking' Fable 5," the company said.
Anthropic added that compliance with the order required immediate action.
"The order means we must abruptly disable Fable 5 and Mythos 5 for our customers to ensure compliance," the company stated.
The shutdown extends beyond the foreign users reportedly targeted by the directive. By Friday evening, access to both models had been suspended entirely. Visitors to Claude's platform were met with a notice stating that "Fable 5 is temporarily unavailable."
According to Axios, a U.S. administration official said concerns emerged after another company reportedly demonstrated an ability to bypass safeguards built into Mythos. The claim reportedly triggered alarm among officials who feared the models could create national-security vulnerabilities if deployed broadly.
Anthropic disputed suggestions that the systems lacked protections. The company said it had conducted extensive testing with government agencies, outside organizations and internal teams before launch.
The company said evaluations showed Fable's safeguards were substantially "more effective" than previous models and that testers had failed to identify a "universal jailbreak" capable of consistently defeating the system's protections.
The Commerce Department's letter reportedly goes beyond a temporary shutdown. Under the directive, licenses would be required for "the export, re-export or domestic transfer of those Anthropic models." The company would also need to submit applications for individually validated licenses, with potential "financial and civil penalties" for noncompliance.
The confrontation comes amid broader tensions between Anthropic and the U.S. defense establishment. In May, the Pentagon announced new agreements with major artificial-intelligence developers to deploy advanced AI capabilities across classified military networks. Anthropic was notably absent from those arrangements.
The company has previously expressed concerns about military applications of artificial intelligence, particularly regarding domestic surveillance and fully autonomous weapons systems. Those positions have occasionally put it at odds with government agencies seeking broader deployment of AI technologies for defense purposes.
Anthropic filed confidential paperwork for an IPO on June 1, earlier than many analysts expected. The company is widely viewed as one of the leading challengers to OpenAI and has attracted significant attention from investors seeking exposure to the rapidly expanding AI sector.
