The White House has imposed export controls on Anthropic, effectively banning access to the company's Fable 5 AI model and a second model called Mythos. The Financial Times described the restrictions as "capricious" — a pointed word suggesting the controls were applied without clear or consistent reasoning.
PCMag reported testing Fable 5 before it "disappeared," indicating the model was briefly available to users before the ban took effect. The publication's framing suggests the removal was abrupt, with little public notice.
The fallout has reached the security community. According to Bank Info Security, cybersecurity leaders are urging the government to restore access to both Fable and Mythos, signaling that the restrictions are disrupting professional work — not just casual experimentation.
The Financial Times' "capricious" label points to a deeper concern: that the White House is deploying export-control mechanisms — tools historically aimed at physical goods like chips and weapons hardware — against AI software in ways that seem inconsistent and hard to predict. Anthropic, the company behind the Claude family of AI models, now finds itself caught between competing government pressures: the administration has simultaneously promoted American AI leadership while restricting one of its most capable domestic models.
If leading AI models can vanish overnight on unclear legal grounds, businesses and researchers that have built workflows around them face serious uncertainty — and that uncertainty may ultimately benefit competitors in countries not subject to U.S. export law.