Anthropic has given the European Union's cybersecurity agency, ENISA, testing access to its Mythos 5 AI model following months of negotiations that began in late May4. The access comes months after the model's initial release1.
ENISA has begun testing Mythos 5 following the green light from Anthropic2. However, the agency still lacks access to Mythos 5.1, the newer version of the model.
ANALYSIS The gap between Mythos 5's release and ENISA's access raises a practical question for EU AI oversight: by the time a regulator secures testing access to one model version, the next iteration may already be in deployment. The absence of Mythos 5.1 access underscores that dynamic.
The arrangement positions ENISA to conduct cybersecurity evaluations of Anthropic's model, though the scope and terms of the testing access are drawn from limited reporting3. Bloomberg's Gian Volpicelli reported on the development, describing Anthropic as having "handed" the EU cybersecurity agency access to the model.