Anthropic has been forced to abruptly disable its most advanced AI models, Fable 5 and Mythos 5, for all users worldwide after the US government issued an export control directive citing national security concerns. The directive, received by the company on Friday, June 12 at 5:21 p.m. ET, ordered the suspension of access to the models for all foreign nationals – including Anthropic’s own employees, regardless of their location. Because the scope of the order made selective compliance impossible, the company said it had no choice but to take both models offline globally. Access to all other Anthropic models remains unaffected.
National security concerns and the ‘jailbreak’ dispute
The US government’s rationale, according to Anthropic, is a belief that a method exists to bypass, or “jailbreak”, the safeguards built into Fable 5 that are designed to prevent its use in identifying software vulnerabilities. The company said it has received only “verbal evidence of a potential narrow, non-universal jailbreak” and that the government did not provide specific details of the national security concern. Anthropic disagrees with the assessment, characterising the technique as a narrow method that exploits minor, previously known vulnerabilities that other publicly available AI models – such as OpenAI’s GPT-5.5 – can also uncover. The company noted that extensive red-teaming exercises conducted with the US government and other third parties had not revealed a universal jailbreak, and that many users had already complained that the safeguards on Fable 5 were “overly broad”.
Anthropic said that the government’s demonstration of the technique revealed only a small number of minor vulnerabilities. “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 said, warning that applying such a standard could halt new model deployments across the industry. The unrestricted version of the model, Mythos 5, was already limited to a vetted group of cyber defenders and critical infrastructure operators; Fable 5 had been designed for wider public availability with additional guardrails. Anthropic believes there has been a “misunderstanding” and said it is working to restore access as soon as possible.
Experts have warned that Mythos models, in the wrong hands, could dramatically accelerate sophisticated cyber-attacks, particularly in sectors such as banking that rely on complex, interconnected and often decades-old technology systems. The risks associated with jailbroken large language models – including data breaches, the manipulation of automated systems and compromised decision-making – have long been flagged by security researchers.
Broader implications for AI regulation and US export controls
The directive marks a significant escalation in US efforts to control foreign access to advanced AI capabilities. For years, American export controls have focused on the chips and tools that power AI rather than on restricting access to the AI models themselves. The Department of Commerce’s Bureau of Industry and Security (BIS) has been updating regulations to impose licensing requirements on advanced computing integrated circuits and AI model weights, but this is the first time a commercially deployed AI model has been directly targeted by an export control order.
The action comes against a backdrop of strained relations between Anthropic and the Trump administration. Earlier this year, the company refused to allow the US military to use its AI models for domestic surveillance and fully autonomous weapons systems. In response, the Pentagon placed Anthropic on a supply chain blacklist, set to take effect later in the year. The Pentagon’s chief information officer, Kirsten Davies, posted on X that the defense department supported prioritising national security. “Some things are simply more important than revenue cycles, clickbait and pre-IPO valuation. America First. Always,” she said.
Anthropic itself has called for greater US oversight of AI, including the ability to block models that pose unacceptable risks. It has proposed a framework requiring mandatory independent safety testing for powerful AI models and granting the government authority to halt deployments that present catastrophic risks. However, the company argued that the directive from the Commerce Department did not follow principles of fair and fact-based regulation. The incident also arrives as Anthropic prepares for a US initial public offering, having confidentially filed with the Securities and Exchange Commission in June 2026. The company’s valuation has reportedly reached approximately $965 billion.
The UK’s Minister for AI and Online Safety has commented on the situation, placing it in the context of technological sovereignty and the UK’s own investment in AI chips. The broader executive order issued by the Trump administration on June 2, 2026, which seeks to balance AI innovation with national security and cybersecurity risks, suggests that more such restrictions may follow. Industry observers noted that if the standard of recalling models over narrow jailbreaks were applied across the sector, it could significantly impede new AI model deployments. Anthropic has said it will share more details within 24 hours of receiving the directive. A US official confirmed that the commerce department had issued the export control directive to suspend all access to Fable 5 and Mythos 5 by foreign nationals.
