Particle.news

U.S. Order Prompts Anthropic to Disable Its Most Powerful AI Models Globally

Washington said the models posed national‑security risks by exposing software vulnerabilities, a decision that has exposed Europe’s dependency on U.S. AI systems and pushed leaders to demand greater digital sovereignty.

Overview

  • U.S. authorities instructed Anthropic to block foreign access to Mythos 5 and Fable 5, and Anthropic responded that it could not restrict users by nationality so it turned the services off worldwide.
  • U.S. officials framed the move as a security step because Mythos can rapidly find and exploit software flaws, while reporting of possible access tied to actors linked to China remains unconfirmed.
  • The shutdown interrupted access for researchers, companies and some government partners and forced CEOs and G7 leaders to raise urgent questions about who controls high‑end AI and how to manage its risks.
  • In Germany recent revelations that politicians used generative AI for speeches and guest pieces — including a removed article after false quotes — have intensified demands for disclosure rules and clearer public‑sector guidelines.
  • Coverage ranges from warnings that the order reveals dangerous concentration of power in a few U.S. firms to critiques that Washington overreached, and the episode is accelerating EU debates on building local compute, stronger oversight and legal remedies.