Anthropic to Open Mythos AI to EU's ENISA - Dark Reading

This arrangement is part of an initiative called Project Glasswing, as reported by Dark Reading. The announcement was made in a statement from the European Commission, though specific details about the scope and duration of access were not disclosed.

Under Project Glasswing, ENISA researchers will be able to test Mythos for security vulnerabilities and share their findings with Anthropic. The model, Mythos, is described as a frontier AI system, meaning it represents one of Anthropic's most advanced capabilities.

Anthropic has previously positioned itself as a safety-conscious AI developer, and this voluntary cooperation with ENISA appears to be part of its engagement with regulators. The initiative underscores growing efforts to subject frontier AI models to external cybersecurity scrutiny.