THE LONG VIEW

Bottom Line Up Front: Anthropic’s restriction of its Mythos and Fable 5 AI systems to only domestic users signals an emerging trend of AI nationalism that could delay or fragment the development of safe, globally aligned AGI while increasing geopolitical AI divides. Threat Identification: The restriction of advanced AI models to specific nationalities represents a shift toward technological protectionism, which may incentivize other organizations or nations to follow suit, leading to fragmented AI development ecosystems [1]. This poses a systemic risk to cooperative AGI safety research and equitable access. Probability Assessment: Given current geopolitical tensions and increasing regulatory scrutiny of frontier AI, it is highly likely (70–80% probability) that more AI labs will implement nationality-based access controls by 2027, especially in the U.S., EU, and China [1]. Impact Analysis: Such restrictions could slow down global AGI progress by limiting diverse input and collaboration, increase the risk of divergent ethical standards, and exacerbate digital divides. Additionally, they may provoke retaliatory restrictions abroad, leading to a bifurcated or multipolar AI landscape with reduced transparency and higher conflict risk. Recommended Actions: 1) Establish multilateral AI access frameworks through forums like the Global Partnership on AI; 2) Develop tiered access models based on security vetting rather than nationality; 3) Increase funding for international AGI safety collaborations; 4) Promote open-weight models with robust usage governance to balance openness and security. Confidence Matrix: Threat Identification – High confidence; Probability Assessment – Medium-High confidence; Impact Analysis – High confidence; Recommended Actions – Medium confidence. [1] The New York Times, "Anthropic Blocks Foreign Nationals From Using Its Mythos and Fable 5 A.I. Systems," published June 13, 2026.