AI-generated·Learn how
© der Standard
Diplomacy·2h ago

U.S. government forces Anthropic to take Fable 5 and Mythos 5 offline globally; Europe sounds alarm on digital sovereignty

The U.S. Department of Commerce ordered Anthropic to suspend access to its most advanced AI models, Fable 5 and Mythos 5, for all foreign nationals, forcing a worldwide shutdown. The move, citing national security concerns, has drawn sharp criticism from European officials and AI industry leaders.

A sudden global blackout

Late on Friday, June 12, Anthropic deactivated its two most advanced AI models, Fable 5 and Mythos 5, across all regions. The San Francisco-based lab acted after receiving a direct order from the U.S. Department of Commerce — issued under the Trump administration — demanding that all foreign nationals be denied access, citing national security. Because the company could not reliably distinguish users by nationality, the only way to comply was to take the services offline entirely. Developer Simon Willison tracked the shutdown in real time: at 18:59 Pacific time, Fable 5 stopped responding and Anthropic began redirecting users to the older Opus 4.8.

Why the government stepped in

The trigger was a report delivered to the administration by an unnamed third party — later identified by several media outlets as Amazon, a major Anthropic investor — claiming to have found a method to bypass the safety guardrails built into Fable 5. These guardrails were designed to prevent the model from being exploited for cyberattacks, including the discovery of software vulnerabilities and the development of chemical or biological weapons. Anthropic called the discovered loophole “narrow” and the exposed vulnerabilities “minor.” The company stressed that it had spent thousands of hours stress-testing its safety measures with government bodies and independent experts before launch. Nevertheless, the administration acted immediately. The relationship between the Trump White House and Anthropic was already strained: the firm had refused to allow the U.S. military to use its models for domestic surveillance or fully autonomous weapon systems, landing it on a government supply-chain blacklist.

European alarm over digital sovereignty

The shutdown sent shockwaves through Europe, where companies, researchers, and public bodies depend on U.S. AI infrastructure. Bitkom President Ralf Wintergerst said the episode has made it “more than clear” that Germany and Europe are at the mercy of Washington’s goodwill when it comes to accessing the strongest AI models. He warned of immediate effects on industrial performance, public administration, security, and scientific excellence. The EU Commission announced it was examining the practical consequences for European users and saw the incident as further proof that Europe must accelerate its technological sovereignty.

Germany and Europe are dependent on the goodwill of the US government for access to the strongest AI models. This has immediate effects on the performance of our classic industries and administrations and also impairs our security and the excellence of our science.

“It changes the rules for the entire industry”

The decision drew sharp criticism from both proponents and skeptics of AI regulation. Entrepreneur Martin Varsavsky said the implications were “enormous,” arguing that any startup working on frontier models is now “at the mercy of the government.” The order, he said, “doesn’t just punish Anthropic. It changes the rules for the entire industry.” Researcher Gary Marcus expressed shock that the administration had undercut U.S. competitiveness: “It didn’t occur to me the Trump administration could trip the US efforts from behind — but it just did.” The move is unprecedented for a democratic government. While China has long embedded restrictions into AI models before release, no U.S. administration had ever forced a domestic company to shut down an already-deployed advanced model.

The order doesn’t just punish Anthropic. It changes the rules for the entire industry.

It didn’t occur to me the Trump administration could trip the US efforts from behind — but it just did.

From launch to blackout: a timeline

The speed of events was striking. Anthropic released Fable 5 on June 10, with CEO Dario Amodei calling publicly for “binding regulation” of frontier AI. By the afternoon of June 12, the Trump administration had issued its sweeping access restriction. That evening, the models disappeared for every user on the planet.

Timeline: from model launch to global shutdown
  1. Anthropic releases Fable 5, a constrained version of Mythos 5; CEO Dario Amodei calls for binding AI regulation.
  2. Trump administration orders Anthropic to suspend access to Fable 5 and Mythos 5 for foreign nationals, citing national security.
  3. Anthropic shuts down Fable 5 globally; developer Simon Willison records the moment the model stops responding.
  4. EU Commission says it is examining practical consequences; Bitkom warns of digital sovereignty risks.
San Francisco · Washington · Brussels

8 sources

Get Pollar Weekly

The week in news, every Friday. Free.

Free. No tracking, no ads. Unsubscribe anytime.

More from Politics & Economy