When Washington Pulls the Plug: What the Anthropic Shutdown Means for Every Tech Company

Last Friday, the U.S. Commerce Department sent Anthropic a letter that effectively forced the AI lab to take its two flagship models, Fable 5 and Mythos 5, offline. The directive invoked an obscure export control rule, barring non-Americans - including Anthropic's own employees - from accessing the models, citing an unspecified national security concern. Anthropic complied immediately, shutting down access for all customers. No court order was involved. The letter itself hasn't been released.

Initial reports suggested the move was tied to a reported jailbreak of Fable 5's guardrails. But new analysis from cybersecurity veteran Katie Moussouris, who reviewed a private paper on the alleged bypass, tells a different story. Moussouris argues that the researchers' technique - asking the model to review code for security issues versus asking it to fix code - should never have triggered an export control. The behavior, she says, "cannot meaningfully be fixed, and any attempt would only weaken the model for defense."

Axios reports that personality clashes between Anthropic and the Trump administration may have played a bigger role than any technical flaw. That raises an uncomfortable question for every tech leader: What happens when a government action is more about politics than policy?

Dozens of top security researchers are now calling on the administration to revoke the order, warning that pulling advanced cybersecurity tools from U.S. defenders is dangerous. The precedent is stark: if the government can shut down an AI model over a misinterpreted research paper, no software company is safe. As one analyst put it, this isn't about a jailbreak - it's about control.

Source: TechCrunch