The European Union Agency for Cybersecurity discusses with Anthropic about the use of the Mythos model for identifying security breaches.

The European Union Agency for Cybersecurity (ENISA) confirms discussions with Anthropic for the use of the AI model Mythos, which can accelerate the discovery of security breaches, but also the associated risks. European Commission officials recently visited San Francisco for negotiations with Anthropic, considering joining Project Glasswing, an American industrial coalition. Although the discussions are productive, the exact conditions for the use of the model have not been established, including the level of access for Anthropic to European systems. The Mythos model was initially limited to companies in critical infrastructure due to its advanced capabilities and the risks of abusive use. Dario Amodei, CEO of Anthropic, emphasized the interest in providing access to allied governments but warned against its use for undemocratic purposes. Anthropic is working on expanding access to Mythos, ensuring that cybersecurity measures are adequate.