Anthropic AI Model Restrictions Prompt G7 AI Security Talks

The U.S. Commerce Department issued an export‑control directive that required Anthropic to suspend access to its frontier models Claude Fable 5 and Claude Mythos 5 for all foreign nationals, including Anthropic’s overseas employees. The company disabled the models for all customers on June 12, 2026, citing national‑security concerns over possible “jailbreak” vulnerabilities.

During the G7 summit in Evian‑les‑Bains, France, the issue dominated a working lunch attended by President Donald Trump, French President Emmanuel Macron, and AI CEOs Sam Altman (OpenAI), Dario Amodei (Anthropic) and Demis Hassabis (Google DeepMind). Altman warned leaders “Do not cede your responsibilities to AI labs like mine,” while Amodei urged democracies to collaborate on AI safety. European Commission President Ursula von der Leyen called for U.S.–EU cooperation to ensure Europe can safely use the best AI models.

The G7 communique also asked finance ministers and central‑bank governors to analyse AI‑related risks for the financial system, and Japanese banking officials warned that AI‑driven cyber‑attacks could disrupt services such as ATMs. Anthropic executives are scheduled to meet U.S. officials to discuss the restrictions, and technology leaders have signed an open letter seeking a more transparent risk‑assessment process.