KI
KIneAngst
All News
🟡 Partially justified

US government forces worldwide shutdown of Anthropic's AI models Fable 5 and Mythos 5 over jailbreak concerns

What it really says

The US Department of Commerce issued an export control directive on June 12, 2026, requiring Anthropic to block access to its latest AI models Fable 5 and Mythos 5 for all foreign nationals worldwide - regardless of whether they are located inside or outside the United States. Even foreign Anthropic employees within the US are affected. The models had been released just three days earlier, on June 9, 2026. The government cites national security interests as justification: there are suspicions that a method exists to bypass Fable 5's safety mechanisms ('jailbreak'). The system could allegedly be misused to discover vulnerabilities in external software and directly exploit them for cyberattacks. Since a technical distinction between US citizens and foreign users could not be implemented on short notice, Anthropic disabled both models for all users worldwide. All other Claude models remain unaffected.

Our assessment

This incident is remarkable from multiple perspectives. On one hand, it demonstrates that governments are willing and able to intervene quickly when concrete security concerns arise - an AI model was shut down worldwide within hours of a government order. For people worried about uncontrolled AI development, this is initially reassuring: functioning control mechanisms exist. On the other hand, the case raises serious questions. According to Anthropic, the order was based only on verbal notice of a 'potential narrow jailbreak' - without detailed technical evidence. Anthropic clearly disputes the proportionality, arguing that by this standard, virtually any new model release could be halted, since jailbreaks are regularly found in all major AI models. The use of export control law rather than conventional product safety regulation is also notable: it implicitly categorizes AI models as security-relevant technologies, comparable to defense goods or encryption technology. For European users, this means concrete dependence on US regulatory decisions - a government in Washington can cut off access to AI tools used productively in Germany from one day to the next.

Relevance for Germany

This incident has immediate implications for Germany and Europe. All users without a US passport - including German companies, developers, and individuals - were cut off from Anthropic's most capable models. Anyone who had already integrated Fable 5 or Mythos 5 into business processes, software development, or research was left without access overnight. The case underscores Europe's structural dependence on US AI providers: the decision about which AI models are available to European users is currently made in Washington, not Brussels. This is likely to further fuel the political debate about European AI sovereignty and building independent AI infrastructure. The Federal Network Agency and the EU AI Office will need to consider how Europe should handle a situation where US export control decisions have direct impact on AI availability in the European single market.

Fact check

The worldwide shutdown of Fable 5 and Mythos 5 is confirmed by Anthropic's official statement dated June 13, 2026. The company published a detailed explanation on its website documenting receipt of the export control directive on June 12 at 5:21 PM ET. Multiple independent media outlets - including CNBC, Bloomberg, Fortune, Heise, and ComputerBase - consistently report the events. The US government's justification (jailbreak concerns and national security) and Anthropic's objection (disproportionate response to a 'potential narrow jailbreak') are consistently reported across all sources. The fact that the models were released on June 9 and blocked just three days later is documented by the official release dates.

Source

  • https://www.anthropic.com/news/fable-mythos-access
  • https://www.cnbc.com/2026/06/12/anthropic-disables-access-to-fable-5-and-mythos-5-to-comply-with-government-directive.html
  • https://www.computerbase.de/news/netzpolitik/anordnung-der-trump-administration-anthropics-top-modelle-fable-5-und-mythos-weltweit-gesperrt.97894/
  • https://www.heise.de/en/news/US-government-forces-shutdown-of-Anthropic-s-AI-Fable-5-and-Mythos-5-11331146.html
Share:
SicherheitRegulierungUSAKI-ModelleMachtkonzentration