US Government Orders Anthropic to Suspend Access to Fable 5 and Mythos 5 Models

US Government Orders Anthropic to Suspend Access to Fable 5 and Mythos 5 Models The sudden shutdown of Anthropic’s most powerful AI models has turned a technical security dispute into a global argument over who controls frontier AI.

On June 9, Anthropic launched Fable 5, a public “Mythos‑class” model built on the same foundation as its tightly restricted cybersecurity engine, Mythos 5. Within days, Fable 5 was topping public benchmarks and “spent three days as the most capable AI model ever released to the public.”

Behind the scenes, however, U.S. officials and investors were growing nervous. IMF chief Kristalina Georgieva had already warned that models like Mythos “can be used to destroy the financial system” if misused. In early June, Amazon security researchers produced a paper claiming they could prompt Fable 5 to reveal information useful for cyberattacks, and CEO Andy Jassy relayed the findings to the White House.

On June 12, after a frantic 24-hour period of calls between senior Trump administration officials and Anthropic CEO Dario Amodei, Commerce Secretary Howard Lutnick sent an export-control directive restricting access to Fable 5 and Mythos 5 for “any foreign national” inside or outside the U.S. Axios reported this marked an escalation in treating cutting-edge AI as a national-security asset. To comply, Anthropic disabled both models globally that night, saying the government had provided only verbal evidence of a “potential narrow, non-universal jailbreak” also achievable with other models. In its own statement, the company stressed it had found only “a small number of previously known, minor vulnerabilities” and argued that “Fable’s safeguards are substantially more effective than those of any previously deployed model.”

The shutdown triggered immediate fallout. Developers lost access to a model that had just “dominated every benchmark” against OpenAI’s GPT 5.5, and India—Anthropic’s second-largest market—saw a new Tata Consultancy partnership thrown into limbo, intensifying calls for a domestic sovereign AI stack. European startup Mistral, long warning that foreign providers “have the keys” to critical models, said the episode validated its push for deployable, open-weight systems.

U.S. allies and rivals drew their own lessons. Canadian Prime Minister Mark Carney likened the concentration of AI dependence revealed by the Anthropic ban to “the 2008 financial crisis,” urging redundancy and diversification in AI infrastructure. Commentators in Europe and Asia argued the move strengthens China’s pitch for its own frontier models and may even be “a gift to China” if partners seek non‑U.S. alternatives.

Inside Washington, Anthropic’s senior technical staff flew to D.C. for crisis talks with the Commerce Department on June 15 to explain their safeguards and seek a way to restore access. Business Insider described the export ban as the most significant test yet of how far the White House will go in blocking commercial AI deployments it sees as dangerous. TechCrunch, citing cybersecurity experts, argued the government’s move “was never about an AI jailbreak” alone but a broader assertion of power over the AI industry.

Whether Fable 5 and Mythos 5 return in their original form now hinges on those negotiations. But globally, the episode has already accelerated a shift toward sovereign and local AI—summed up in one viral reaction: “Fable is banned. Long live local AI.”

Continue reading https://foxvector.com/stories/019ed0cb-e275-3dc3-70d9-3ac8f2505439