The US government imposed export controls on June 12 restricting foreign governments, companies, individuals, and foreign-born employees from accessing Anthropic’s latest AI models, Mythos 5 and Fable 5, citing national security concerns [1, 2, 3]. To comply, Anthropic disabled access to these models for all users worldwide the same day [4, 5, 3].
The US Commerce Department, led by Secretary Howard Lutnick, issued a letter to Anthropic CEO Dario Amodei specifying the export restrictions and warning of penalties for any violations [1, 6, 7]. Lutnick’s letter stated, “Until further notice, you must submit an application for an individually-validated license prior to the export, re-export, or transfer... of the Mythos or Fable models to any destination worldwide or to any ‘foreign person’ wherever located” [6]. The export controls apply globally, including to foreign nationals inside the US and foreign-born employees of Anthropic [1, 8].
US officials cited a possible "jailbreak" or safeguard bypass vulnerability in Mythos 5 and Fable 5 that could allow misuse, including identifying software flaws [1, 9]. The US government also suspected that a China-affiliated group had accessed the models, heightening national security concerns [9]. In response, Anthropic disputed the severity of the jailbreak finding, saying it is narrow, limited to minor known issues also found in other AI models, and does not justify the ban. 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” [4, 10]. On social media, Anthropic stated it believes the export controls represent a misunderstanding and it is working to restore access as soon as possible [3].
Anthropic’s relationship with the US government had already frayed earlier in 2026 after the company refused military uses of its AI, leading to a Pentagon blacklist [4, 7]. Meanwhile, Anthropic confidentially filed for a US initial public offering in May 2026 with a valuation exceeding $900 billion [11, 8]. Experts such as security analyst Katie Moussouris argued the reported AI guardrail bypass does not warrant export controls and suggested political factors may influence the US government’s position [10].
The US actions mark a significant escalation in export controls by targeting finished AI models rather than hardware or software tools alone [11, 9]. Negotiations between Anthropic and Commerce Department officials have been ongoing, including meetings on June 15 involving Secretary Lutnick and National Cyber Director Sean Cairncross [2, 8]. The export controls remain in place until the US government's national security concerns are resolved, which is expected possibly in the coming weeks [1, 7].
