Overview
- The U.S. government instructed Anthropic to block access to its newest models for non‑U.S. users, and Anthropic disabled those models globally because it could not technically restrict users by nationality.
- U.S. officials cited a reported jailbreak prompt sequence and concerns about foreign access to one model as the reasons for the order, though independent confirmation of the alleged Chinese‑linked access and the full technical details remain limited.
- The sudden removal disrupted businesses and projects that had begun relying on Anthropic’s Fable and Mythos tools, and it complicates the company’s commercial plans and investor messaging.
- European leaders and industry groups have pushed for faster investment in sovereign AI capacity, with plans discussed for a Franco‑German R&D centre and calls to expand chips, data centres and domestic compute capacity.
- The episode highlighted a structural risk: Europe controls only a small share of high‑performance AI compute and depends on U.S. providers, a shortfall that experts say makes the continent vulnerable to future access restrictions.