US Government Orders Shutdown of Anthropic’s Fable 5 and Mythos 5 AI Models The abrupt shutdown of Anthropic’s Fable 5 and Mythos 5 models has turned a technical dispute over an alleged “jailbreak” into a global fight over who controls advanced AI.

On June 9, Anthropic publicly launched Fable 5, a guarded version of its Mythos 5 cybersecurity model, which it had previously deemed too powerful for general release. Three days later, Amazon security researchers privately warned the White House they could prompt Fable to reveal code vulnerabilities, triggering national security alarms inside the Trump administration. On June 12, Commerce Secretary Howard Lutnick sent an export-control letter barring access by “any foreign national,” including Anthropic’s own staff, forcing the company to disable both models worldwide to comply.

Anthropic publicly pushed back, calling the move disproportionate. The firm said it had only seen a “narrow, non-universal jailbreak” that surfaced “relatively simple” flaws also discoverable with other public models, and warned that if this standard were applied “it would essentially halt all new model deployments for all frontier model providers.” In a detailed statement, the company stressed that no one had found a universal jailbreak and that its safeguards were “substantially more effective than those of any previously deployed model.”

Inside Washington, officials argued they had little choice. An Axios scoop described the export controls as part of an effort to treat cutting-edge systems as “national security assets,” with licenses now required for any transfer of Mythos 5 or Fable 5. Business Insider later reported tense calls and a 90‑minute ultimatum to Anthropic’s CEO before the order went out, after Amazon CEO Andy Jassy raised concerns directly with senior officials. Some reports also cited fears that China-linked actors might gain access and distill the model, though Anthropic said Beijing was not raised in talks.

As access vanished, the backlash widened. Cybersecurity leaders organized an open letter arguing that yanking “the best models away from defenders” while adversaries keep building “is not safety, it is sabotage,” and warning the ban “risked America’s AI leadership without any real risk to justify it.” Alex Stamos, former Facebook security chief, told Axios the move set a precedent that “American models can’t do defensive security research,” potentially weakening US cyber defenses.

Abroad, governments and rivals seized on the moment. India’s tech community treated the suspension—which cut off Anthropic’s second‑largest market overnight—as a “warning shot” about relying on foreign AI, turbocharging calls for sovereign infrastructure and a multibillion‑dollar domestic AI fund. European startup Mistral said the episode proved that “at some point, you need to be able to turn it off or turn it on, and you don’t want to leave it to another country,” reinforcing its pitch for open‑weight, self‑hosted models.

Allies are now scrambling to contain the fallout. US and European officials are discussing a “trusted partner” scheme to give close allies controlled access to frontier models after the Anthropic dispute, while Canadian Prime Minister Mark Carney compared the shutdown’s systemic risk to the 2008 financial crisis and urged diversification of AI infrastructure.

Back in Washington, Anthropic has flown senior technical staff to D.C. for crisis talks with the Commerce Department and the National Cyber Director, seeking a truce that could restore Fable and Mythos under revised safeguards. But as one FT editorial warned, the White House’s “opaque” and “capricious” controls risk chilling innovation across an industry whose sky‑high valuations depend on global access to their most powerful models.

Continue reading https://foxvector.com/stories/019ed403-7334-05e8-7141-3619fc052405