US Government Orders Anthropic to Suspend Access to Fable 5 and Mythos 5 AI Models The U.S. government’s sudden shutdown of Anthropic’s most powerful AI systems has turned a narrow technical dispute into a global fight over who controls frontier AI.

On June 9, Anthropic released Fable 5 to the public and Mythos 5 to vetted partners, describing Fable as its most capable generally available model and Mythos as the same system “with the safeguards lifted in some areas.” Just days later, Amazon security researchers told the White House they had prompted Fable 5 to reveal information that could aid cyberattacks, a finding CEO Andy Jassy escalated to senior officials.

By the afternoon of June 12, Commerce Secretary Howard Lutnick had sent Anthropic an export-control order barring access to Fable 5 and Mythos 5 by “any foreign national,” including the company’s own staff. Anthropic says the letter cited national security law but “did not provide specific details” and that the concern appears to be a “potential narrow, non-universal jailbreak” that only surfaces minor, already-known vulnerabilities also discoverable by other public models. To comply, the company disabled both models worldwide that evening.

Officials argue they acted to prevent misuse of a cybersecurity-focused system that could help exploit critical infrastructure, and say the models must stay “locked down until the U.S. government’s national security apparatus is hardened.” Some in Washington reportedly viewed Anthropic as a “bad actor” whose handling of vulnerabilities showed “recklessness.”

Anthropic, already in a separate fight over being labeled a Pentagon “supply chain risk,” has pushed back, warning that if such a standard were applied broadly it would “essentially halt all new model deployments for all frontier model providers.” The company has flown senior technical staff to Washington for crisis talks, calling the move a “misunderstanding” it is working to resolve.

Abroad, the shutdown is feeding calls for “sovereign AI.” In India, the loss of access to Anthropic’s top models days after a major Tata Consultancy Services partnership has intensified debate over whether the country can afford to rely on U.S.-governed systems. European startup Mistral says the saga validates its push for independent, open-weight models, warning that countries “don’t want to leave it to another country” to decide when AI can be turned on or off.

Even U.S. allies are uneasy. Canadian Prime Minister Mark Carney likened the systemic risk of concentrating AI in a few models to the 2008 financial crisis and urged governments to “diversify” and build redundancy into AI infrastructure.

Within the industry, the episode is also boosting interest in local and open models. One viral X post captured the mood: “Fable is banned. Long live local AI,” urging developers to learn to run powerful systems on their own hardware.

Continue reading https://foxvector.com/stories/019ed026-adc4-28fe-7014-35ab0ff787c1