A recent US government decision to impose immediate export controls on Anthropic’s AI models Fable 5 and Mythos 5 has sparked widespread industry scrutiny over regulatory predictability and AI safety governance. The crisis unfolded on a recent Friday in Washington, D.C., when Treasury officials, alerted by internal Amazon research identifying a potential model jailbreak, issued a 90-minute ultimatum to Anthropic. The directive mandated strict restrictions on foreign national access to both Fable 5, its public-facing variant, and the underlying enterprise-grade Mythos 5 framework. Unable to implement the citizenship-based access controls within the compressed timeframe, Anthropic temporarily suspended global access to both models. Anthropic originally positioned Mythos 5 as a high-risk system requiring robust safeguards, launching Fable 5 with extensive security guardrails. Despite independent red-teaming suggesting the safeguards held against early exploitation attempts, the administration treated the findings as a critical security breach. Cybersecurity leaders and technology executives widely criticized the response as abrupt and poorly coordinated, noting that the exploited capabilities are replicable across multiple frontier models. Industry observers emphasize that the administration’s reaction appeared accelerated and uncompromising, particularly given Anthropic’s ongoing legal and policy disputes with federal agencies over military AI usage. The incident highlights significant tensions in the current regulatory environment. While the administration previously discarded voluntary safety review frameworks in favor of a lighter touch, this crisis demonstrates how ad-hoc export controls can rapidly dictate model deployment and compliance strategies. Tech leaders warn that unpredictable executive mandates risk undermining the United States’ strategic objective of maintaining AI leadership, prompting some enterprises to accelerate plans for alternative hardware and non-US model partnerships. Furthermore, the administration’s shifting justification from foreign access concerns to potential executive order non-compliance has fueled perceptions of disjointed policy-making. Legal and industry sources anticipate a likely rollback of the blanket foreign-national restriction amid intense corporate pushback, though the broader regulatory approach remains unsettled. The episode underscores the absence of a stable, rules-based governance structure for frontier AI development. Stakeholders across government and industry are increasingly advocating for a transparent, predictable compliance framework to manage high-risk model deployment, rather than relying on crisis-driven executive mandates. As negotiations continue, the resolution of this dispute will significantly influence how American AI firms navigate international competition, safety standards, and federal oversight in the coming months.