U.S. Commerce Department Blocks Fable 5 Access Amid AI Security Concerns

According to Beating Monitor, the U.S. government’s emergency suspension of access to Anthropic’s Fable 5 and Mythos 5 was the culmination of weeks of behind-the-scenes maneuvering. According to informed sources, even before the two models were officially released on June 9, the U.S. Department of Commerce, led by Secretary Howard Lutnick, had expressed serious concerns about the models’ exceptional cybersecurity reasoning capabilities and had privately attempted to intervene and block their release—without success. To mitigate potential security risks, Anthropic adopted a compromise: it released a commercial version of Fable 5 with extremely strict filters to the public, while restricting the unfiltered “naked” model, Mythos 5, under tight controls within “Project Glasswing,” granting access only to a select few red teams and defense agencies that passed rigorous background checks. However, just three days after the models’ release, third-party entities claimed to have successfully “jailbroken” the controlled Mythos 5 and reported their findings. This immediately provided Lutnick and the Department of Commerce with the perfect opportunity to intervene. The department not only issued a formal letter to Anthropic overnight but also unprecedentedly froze the commercial access rights to the specific models using export control measures. Industry observers widely note that the Department of Commerce’s direct application of export controls to the models themselves sets an extremely dangerous regulatory precedent and signals that the U.S. government is shifting toward a unilateral, forceful approach of reclaiming and interfering in AI safety oversight.