What Europe Should Learn From the Claude Shutdown, According to Germany’s c’t 3003

At the opening, Germany’s well-known tech outlet c’t 3003 bluntly declares that "Claude Fable, or Mythos, is over." The reason, it says, is that the U.S. government issued an order banning non-U.S. nationals, including Anthropic employees, from using the AI models.

In practice, it is impossible for the server side to perfectly distinguish who is and is not a U.S. national. Anthropic therefore shut the models down entirely. According to the program, the U.S. government publicly justified the move by pointing to jailbreaks and security holes.

Anthropic’s rebuttal is straightforward: "OpenAI’s GPT-5 can do the same thing, so Anthropic is not uniquely dangerous." c’t 3003 suggests there may be a political subtext behind that dispute. Unlike some rival AI companies, Anthropic has refused to allow its models to be used in autonomous weapons systems, and the segment argues that the shutdown order can be read as a political penalty for that stance.

In other words, the show suggests this was not purely a technical safety issue. It may also have carried a punitive element directed at a company that refused to align itself with the defense industry.

Toward the end, the reporter goes further and says that, from a European perspective, the episode feels even more threatening. As organizations and individuals across Europe and the wider world grow more dependent on U.S.-built AI services, this incident makes one risk concrete for the first time: a single decision by the U.S. government can cause the service itself to disappear immediately.

The segment closes with the metaphor of "how easily the water tap can be turned off." If AI is becoming part of social infrastructure, then the question of which country controls it, and who can shut it off for political reasons, becomes a direct question of European technological sovereignty.