EU Lobbies U.S. for Access to Anthropic Mythos AI

The European Commission is lobbying Anthropic PBC and the United States government to secure access to Mythos, an AI model capable of identifying deep software vulnerabilities and generating exploits. Commission officials traveled to San Francisco to negotiate access for European banks and tech firms, seeking to use the tool to protect critical infrastructure from network flaws.

Anthropic manages the model through Project Glasswing, currently limiting access to a small group of Wall Street firms and the National Security Agency. While the company intends to release Mythos-class models to all customers after implementing stronger safeguards, it informed the European Union that any expanded distribution requires U.S. government approval. The White House has previously blocked Anthropic's attempts to broaden access, citing security concerns.

U.S. Treasury Secretary Scott Bessent stated the administration is balancing innovation with safety to maintain a strategic lead over China. Meanwhile, Anthropic CEO Dario Amodei warned of a narrow window to patch vulnerabilities before Chinese AI develops similar capabilities. Eurogroup President Kyriakos Pierrakakis emphasized that AI necessitates international governance frameworks and a commitment to maintaining communication channels with the U.S.