US orders Anthropic to disable AI models for all foreign nationals

The U.S. government demanded that Anthropic disable access to its latest AI models, Fable 5 and Mythos 5, for all foreign nationals, including those in the U.S., citing national security risks. The order, received at 5:21pm (21:21 GMT) on June 12, required immediate action, leaving the company no time to prepare. Mythos 5, a non-public model used by government agencies and select corporate partners, can identify decades-old software vulnerabilities, raising fears it could be exploited as a cyberweapon. Anthropic emphasized it received only partial details about the security concerns, but its experts linked the ban to Mythos 5’s ability to review and correct program code. The company argued that similar capabilities exist in rival models like OpenAI’s GPT-5.5 and that Fable 5’s safety measures were rigorously tested. Despite this, the U.S. government insisted on the restriction, affecting hundreds of millions of users globally. Mythos 5’s full capabilities remain restricted to approved users, while Fable 5, released this week, retains its core AI functions but with blocked cybersecurity and biotechnology features. Anthropic’s compliance reflects broader tensions over AI development, especially after the company’s June proposal to pause advanced AI systems globally, warning of an uncontrollable technology race. The move underscores growing regulatory scrutiny over AI’s dual-use potential, particularly in cybersecurity and biotechnology. While Anthropic disputes the necessity of a blanket ban, the U.S. government’s action signals heightened concerns over foreign access to cutting-edge AI tools. The company continues to cooperate but has criticized the lack of transparency in the decision-making process.