US Government Orders Anthropic to Suspend Access to Advanced AI Models Anthropic’s most powerful AI systems, Fable 5 and Mythos 5, went from record‑breaking debuts to a full global shutdown in just three days, exposing a deep rift between Washington and Silicon Valley over how to police frontier models.

On June 9, Anthropic launched Fable 5 to the public, built on its restricted Mythos 5 cybersecurity model. Early benchmarks showed Fable 5 “dominated every benchmark” against OpenAI’s GPT‑5.5, briefly becoming “the most capable AI model ever released to the public.” Mythos 5, meanwhile, was kept to a vetted defensive‑security program called Project Glasswing.

Within days, Amazon researchers reported a “fix this code” technique that could coax Fable 5 into surfacing software vulnerabilities, bypassing its safety refusals by rephrasing prompts. Officials worried this amounted to a jailbreak exposing “automated hacking capabilities” that might escape “regulatory boundaries.”

On June 12, Commerce Secretary Howard Lutnick sent Anthropic an export‑control letter placing Mythos 5 and Fable 5 under a licensing regime for any foreign access, including “foreign national Anthropic employees.” Axios reported this as the Trump administration “blocking foreign governments, companies and individuals from accessing Anthropic’s most advanced AI models,” escalating efforts to treat cutting‑edge systems as national‑security assets.

Anthropic responded that the only practical way to comply on short notice was to “disable Fable 5 and Mythos 5 for all our customers,” globally suspending what it had billed as “the world’s most powerful cybersecurity model.” The company said it had received only “verbal evidence of a potential narrow, non‑universal jailbreak” and “disagreed that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people.”

From the government’s side, officials argued the models must “remain locked down until the U.S. government’s national security apparatus is hardened,” with some privately describing Anthropic’s handling of known vulnerabilities as “recklessness” that damaged trust.

Cybersecurity experts pushed back. An open letter signed by roughly 100 practitioners warned that “this action has taken the best models away from defenders, created market uncertainty, and risked America’s AI leadership without any real risk to justify it,” calling the move “sabotage” rather than safety. Researchers noted the exploit “is not a jailbreak” and that other leading models such as GPT‑5.5 can surface similar bugs without facing bans.

As Anthropic executives flew to Washington for crisis talks with Commerce and White House cyber officials, analysts warned the episode shows the AI industry is “not immune from U.S. government interference” and that “dangerous” dual‑use models will emerge from multiple labs regardless of this specific shutdown. For now, the world’s top‑performing model remains offline—less because of what it can do technically than because of what it has come to represent in an intensifying power struggle over AI.

Continue reading https://foxvector.com/stories/019eddac-27ce-1ec5-71ba-19b0f0f02bf1