Anthropic is disabling its Fable 5 and Mythos 5 AI models for all users following a US government order citing national security risks. The directive requires the company to suspend access for foreign nationals, though specific concerns were not detailed. Anthropic believes the government fears a potential “jailbreak” allowing the models to be used for identifying software vulnerabilities. This action follows earlier tensions with the US government stemming from Anthropic’s refusal to allow military use of its AI for surveillance and autonomous weapons. The move represents a significant escalation in US efforts to control the AI capabilities of potential adversaries, shifting focus from hardware to access restrictions. Anthropic disputes the justification for recalling the models based on a “narrow potential jailbreak.” The situation highlights growing conflict between AI developers and regulators regarding the assessment and mitigation of AI safety risks.