The U.S. Commerce Department invoked an obscure export control directive forcing Anthropic to pull its Fable 5 and Mythos 5 AI models offline, citing an unspecified national security concern. New details suggest the action was not driven by a genuine AI jailbreak threat but by political tensions between Anthropic and the Trump administration. Cybersecurity expert Katie Moussouris reviewed the underlying research paper and concluded the alleged guardrail bypass was trivial and should never have triggered export controls, and that fixing it would actually weaken defensive capabilities. Dozens of security researchers have called on the administration to revoke the order. The incident sets a precedent for unilateral government shutdown of AI products without court approval, raising concerns about political interference in the U.S. tech industry.
