Companies·3 min read·Anthropic

Anthropic Pulls Fable 5 and Mythos 5 Worldwide to Comply With a U.S. Export-Control Directive

Three days after launch, Anthropic disabled Claude Fable 5 and Mythos 5 for every customer to comply with a Commerce Department order that bars foreign-national access on national-security grounds. The company calls it a misunderstanding over a "minor" jailbreak and is fighting to restore access.

U.S. EXPORT-CONTROL DIRECTIVE Fable 5 & Mythos 5 Pulled Offline Anthropic complies, disabling both for everyone BITSMINDS.COM
Share:

Anthropic has abruptly disabled Claude Fable 5 and Claude Mythos 5 for every customer worldwide, complying with a U.S. government directive that orders the company to cut off access to the two models. In a public statement, Anthropic said it received the order on June 12 at 5:21pm ET — just three days after it launched the pair as its most capable models yet — and that it had no choice but to take both offline immediately to comply.

According to the company and to reporting from CNBC and Bloomberg, the directive came from the U.S. Department of Commerce, in a letter from Secretary Howard Lutnick to chief executive Dario Amodei. Invoking export-control and national-security authorities, the order requires a license for the export, re-export, or even domestic transfer of the models, and bars access by any foreign national — whether inside or outside the United States, and explicitly including Anthropic's own foreign-national employees. Because policing that line user-by-user in real time is impractical, the company says the only way to comply was to switch the models off for all of its hundreds of millions of users at once. Existing sessions now error out; new sessions fall back to other Claude models.

The stated trigger is a security concern. Anthropic says the government believes it has identified a way to bypass, or "jailbreak," Fable 5; per Axios, the Commerce Department acted after another company claimed it had jailbroken Mythos. The specific demonstration, Anthropic says, involved asking the model to read a particular codebase and fix its software flaws. The company reviewed the demo and characterized it as "a small number of previously known, minor vulnerabilities," adding that comparable capabilities are already available in rival models such as GPT-5.5.

Anthropic is pushing back. "We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people," the company wrote, calling the episode a misunderstanding and saying it is working to restore access as soon as possible. No timeline was given. Crucially, the order is narrow in one respect: access to all of Anthropic's other models is unaffected.

The timing is jarring. Fable 5 and Mythos 5 arrived on June 9 to a wave of state-of-the-art benchmark claims, and Fable 5 in particular was marketed on its sanded-down cyber edges — a model deliberately tuned to be harder to misuse. Three days later, a contested jailbreak finding has taken the flagship of that safety-first pitch off the market entirely. Whatever the merits of the underlying vulnerability, the precedent is the story: a single government letter, sent after close of business, can dark a frontier model for the entire planet overnight.

Want AI news before everyone else?

The morning's most important AI stories, straight to your inbox. No fluff.

Related Articles

NVIDIA × OPENAI · REPORTED TALKS Who Vouches for OpenAI? guarantees the debt …which buys its chips NVIDIA OpenAI The answer is the company selling it the chips
Companies

Nvidia May Guarantee $250 Billion So OpenAI Can Buy Nvidia Chips

FAUNA SYSTEMS · LIVING ROBOTS Xenobots, Now as a Service AI-designed frog-cell swarms built to sniff out water pollution PFAS frog stem cells, no edits AI-evolved body shape detects forever chemicals Co-founded by original Xenobot researchers Michael Levin & Josh Bongard · targeting aquaculture & wastewater
Companies

Xenobots Get a Business Plan: Fauna Systems Is Commercializing Living Robots Designed by AI

AMD × ANTHROPIC · ADVANCING AI 2026 2 gigawatts, and $5 billion AMD bets big on Anthropic — and takes direct aim at Nvidia AMD Instinct MI450 Anthropic · Claude 2 GW of MI450 $5B AMD equity stake first GW · H1 2027 Helios rack-scale systems · Claude will help tune AMD’s ROCm · a rare crack in Nvidia’s grip
Companies

AMD and Anthropic Strike a $5 Billion, 2-Gigawatt Deal — a Rare Crack in Nvidia's Grip