Anthropic to expand Mythos AI access in the EU

Anthropic is preparing to give European Union access to its most advanced AI model, Mythos, after weeks of discussions with EU officials. The European Commission said the talks have been productive and that it wants a better understanding of the model’s cybersecurity risks. Mythos was first released to a limited set of companies in April as part of Anthropic’s cybersecurity effort, Project Glasswing. The model was praised for finding software weaknesses, but it also raised concerns that it could be misused by cybercriminals. EU officials said they are continuing discussions with the United States and other partners as powerful AI systems become more widely available. The exact terms of Anthropic’s arrangement with the bloc have not been disclosed.