EU Gains Access to Anthropic's Mythos 5 Cyber AI for Testing

• •ENISA has begun testing Anthropic's Mythos 5 after gaining access on September 11, 2026.
• •Access to Mythos 5 was delayed due to U.S. export restrictions and negotiations with Anthropic.
• •The EU's ability to independently evaluate advanced AI models is critical for cybersecurity assessments.

The EU's cybersecurity agency ENISA has secured access to Anthropic's Mythos 5, an advanced AI model designed to identify and exploit software vulnerabilities. This access comes after months of negotiations and is crucial for independent testing, as the model was previously restricted due to its capabilities. Mythos 5 was initially released to a select group of around 200 organizations under Project Glasswing, which faced complications from U.S. export controls. The European Commission's spokesperson confirmed the testing is now underway, although Anthropic has since released an updated version, Mythos 5.1, which remains inaccessible to EU institutions. The situation highlights ongoing geopolitical tensions regarding AI technology access between the U.S. and Europe.