Anthropic to Give EU Access to Mythos AI Model

Anthropic will share access to its most advanced AI model, Mythos, with the European Union after the region requested permission over cybersecurity concerns. The European Commission confirmed to CNBC on Monday that it had held several productive meetings with the American AI firm. EU tech sovereignty spokesperson Thomas Regnier said the bloc aims to assess potential risks the technology poses.

Anthropic initially released Mythos in April to a limited number of companies under its cybersecurity venture Project Glasswing. The model is designed to identify security flaws in software. The launch prompted concern that the technology could be misused by bad actors to accelerate cybercrime.

Governments, banks, and tech firms noted that Mythos can uncover thousands of previously unknown software vulnerabilities.

The EU had already received access to OpenAI's GPT-5.5-Cyber model in May. The Commission said it held four or five meetings with Anthropic and that agreements were at a different stage. The Trump administration announced separate agreements in May with Google DeepMind, Microsoft, and xAI that allow the government to assess AI models before public release.

Official terms and conditions for the EU-Anthropic arrangement remain unclear. Anthropic did not immediately respond to CNBC's request for comment.