EU cybersecurity agency begins testing Anthropic, OpenAI AI models

The European Union's cybersecurity agency ENISA has been granted access to Anthropic's Mythos 5 artificial intelligence model and is now testing it, the European Commission confirmed on Thursday, more than three months after the company first signaled the bloc would receive access to the system.bloomberg+1

ENISA has also been granted access to OpenAI's latest model, GPT-6-Astra, a Commission spokesperson said.reuters

A Long-Awaited Handover

"Following our constructive engagement with Anthropic, we can confirm that the EU's cybersecurity agency ENISA has been granted access to Mythos 5 and is testing it now," Thomas Regnier, a spokesperson for the European Commission, said in an email. The access comes after months of scrutiny over the delay between Mythos 5's release and the point at which EU regulators could examine it — a gap that critics said undermined the bloc's ability to provide timely oversight of powerful AI systems.bloomberg

Broader Regulatory Push

The dual access grants to both Anthropic's and OpenAI's frontier models reflect the EU's expanding effort to bring advanced AI systems under regulatory review. Under the EU AI Act, which began phasing in its requirements, providers of general-purpose AI models face transparency and risk-assessment obligations. Granting ENISA direct access to test these models is part of a broader push by Brussels to ensure that the most capable AI systems meet safety and security standards before — or shortly after — reaching European users.reuters+1

What Comes Next

ENISA's testing of both Mythos 5 and GPT-6-Astra is now underway, though neither the Commission nor the companies have disclosed the specific scope or timeline of the evaluations. The delayed handover has already fueled debate among European lawmakers about whether voluntary cooperation from AI developers is sufficient, or whether binding deadlines for regulatory access should accompany future model releases.bloomberg+1