Anthropic Grants EU Cybersecurity Agency Access to Mythos Model

Anthropic has provided the European Union's Network and Information Security Agency (ENISA) with access to its Mythos 5 large language model, according to statements from EU officials. The development comes months after the model's initial release and reflects ongoing dialogue between the technology company and European regulators seeking to understand advanced AI systems.

Thomas Regnier, speaking on behalf of EU authorities, confirmed that ENISA is currently conducting testing of the Mythos 5 model. The access grant follows what officials described as "constructive engagement" between Anthropic and the EU's regulatory bodies.

The move represents part of broader efforts by European authorities to evaluate frontier AI models as the region implements its AI Act framework. Regulatory agencies across the EU have been working to assess the capabilities and potential risks associated with advanced AI systems developed by major technology companies.

ENISA's testing of Mythos 5 is expected to contribute to the EU's understanding of the model's technical characteristics and security implications. The agency's findings could inform future regulatory approaches to AI oversight in Europe, particularly concerning cybersecurity considerations.