Definition
AI sovereignty refers to a nation’s or region’s ability to access, develop, and govern frontier AI capabilities without dependence on foreign policy decisions — particularly access to models, compute, and data residency under local jurisdiction.
2026 Context
US export controls on anthropic Mythos 5 and Fable 5 (June 12, 2026) accelerated European AI sovereignty discourse. Key distinction: buying access vs guaranteeing access under local legal frameworks.
Policy Responses
- Austria (June 28): State Secretary Alexander Pröll urged EU to explore hosting anthropic within the bloc (2026-06-28-austria-eu-anthropic-hosting-sovereignty)
- EU Commission: Proposed laws boosting domestic cloud, AI, and semiconductor industries
- France: Gabriel Attal likened Anthropic shutdown to Strait of Hormuz blockade
Industrial / Physical AI Sovereignty
- Japan (July 2026): frontia-project via noetra — own multimodal robotics models on domestic data; compute is nvidia-rubin (sovereignty = models/deployment, silicon rented) (2026-07-16-japan-noetra-tnw-rubin-factory)
Product Responses
- sakana-ai fugu: multi-agent orchestration routing around restricted models
- 360-security Tulongfeng: domestic Mythos-equivalent (vendor claims unverified)