U.S. orders Anthropic to disable Fable 5 and Mythos 5 worldwide over 'fix this code' jailbreak

A Commerce Department export-control directive issued June 12 forced Anthropic to pull its two most advanced models for every customer; about 100 cybersecurity professionals have signed an open letter calling the move counterproductive.

At 5:21 p.m. ET on June 12, a Commerce Department export-control directive landed at Anthropic instructing the company to cut off access to Fable 5 and Mythos 5 for every user “whether inside or outside the United States, including foreign national Anthropic employees.” By Friday the models were dark, worldwide, with no carve-outs for the critical-infrastructure operators who’d been running Mythos 5 through Project Glasswing, Anthropic’s restricted trusted-access program for cyber-defenders.

The trigger was three words. Amazon researchers had demonstrated a “fix this code” jailbreak against Fable 5, and Amazon CEO Andy Jassy escalated the finding directly to the White House by phone. A consumer-cloud CEO calling the executive branch to flag a competitor’s safety posture is, in itself, a story about how AI policy now gets made: through bilateral phone calls between platform incumbents and the agencies that can throttle their rivals.

Anthropic’s read of the same finding is sharply different. The company says its review surfaced “a small number of previously known, minor vulnerabilities” that “other publicly-available models are able to discover,” naming OpenAI’s GPT-5.5 as one. Fable, it noted, was red-teamed for thousands of hours with the U.S. government, the UK AI Security Institute, and outside firms before launch. Mythos was the first model to clear both UK AISI cybersecurity test ranges.

The defender community has lined up behind that framing. An open letter organized by Alex Stamos, chief security officer of Corridor, and signed by roughly 100 cybersecurity professionals from Nvidia, Adobe, Google, Zoom, and Sophos, calls for the controls to be rescinded. Katie Moussouris, founder of Luta Security, laid out the technical case in the post that surfaced the jailbreak’s origin. “To pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous,” the letter reads.

The backdrop sharpens the politics. The Defense Department earlier designated Anthropic a supply-chain risk after negotiations collapsed, and Anthropic has sued to reverse the label; that litigation is ongoing. Export-control authority, historically wielded against foreign adversaries, is now being aimed at an American lab’s flagship product on the basis of a vulnerability class its competitors share.

U.S. orders Anthropic to disable Fable 5 and Mythos 5 worldwide over 'fix this code' jailbreak

Sources