Anthropic is in discussions with the European Union’s cybersecurity agency, ENISA, to provide access to its artificial intelligence-driven cybersecurity system, Mythos. If finalized, this agreement would grant ENISA access to the tool for the first time outside the United States and the United Kingdom, signaling a significant expansion of Anthropic’s cybersecurity collaboration with international partners.
The move would integrate ENISA into Anthropic’s broader Project Glasswing, a platform that allows participating organizations to detect vulnerabilities in their systems using AI. Access to Mythos has been limited so far due to concerns over the technology’s potential misuse in cyberattacks. ENISA’s involvement reflects growing recognition within the EU of AI’s role in fortifying digital security infrastructures.
Project Glasswing launched earlier this year and has attracted major technology and cybersecurity players, including Amazon, Apple, Microsoft, CrowdStrike, Google, Nvidia, and Palo Alto Networks. The initiative focuses on using AI to identify and mitigate high-severity security flaws in critical software systems worldwide. To date, Mythos has uncovered over 10,000 significant vulnerabilities, with partners reporting more than a tenfold increase in bug detection rates.
Anthropic aims to extend Project Glasswing’s reach by onboarding more partners, including governments aligned with the United States, to enhance collective cybersecurity defenses. The company has stressed the importance of developing stronger safeguards before releasing Mythos-class AI models for general use, underscoring a cautious approach to deploying powerful AI technologies. These steps indicate a strategic effort to balance innovation with security concerns as AI tools become increasingly integrated into digital defense frameworks.