Anthropic grants EU cybersecurity agency access to AI security model

Anthropic has granted the European Union Agency for Cybersecurity (ENISA) access to its restricted AI security model, Mythos, through the Project Glasswing program. This initiative allows ENISA to test the model's vulnerability-finding capabilities before a wider release. The collaboration aims to enhance cybersecurity defenses across the EU by leveraging advanced AI to identify and mitigate potential threats. Anthropic's decision to involve ENISA underscores the importance of international cooperation in AI safety and security. The model, designed to detect vulnerabilities in code and systems, represents a significant step in proactive cybersecurity measures. Project Glasswing serves as a controlled environment for rigorous testing, ensuring the model meets high standards of reliability and effectiveness. This partnership marks a milestone in the integration of AI into critical infrastructure protection, with potential implications for global cybersecurity practices.