The European Union’s cybersecurity agency ENISA has been granted access to Anthropic’s Mythos 5 model and is now testing it, a European Commission spokesman told Reuters on September 10. The same spokesman said ENISA has also been granted access to OpenAI’s latest model, GPT-6 Astra.
The confirmation lands after months of public debate over how European authorities would evaluate frontier models with strong cyber capabilities. Anthropic has described Mythos as unusually strong at finding and exploiting security vulnerabilities; OpenAI’s Astra rollout earlier this month carried its own cyber-risk warnings. Reuters’ Thursday wire does not spell out which Mythos build ENISA received, what safeguards apply, or how long the tests will run.
Reporting was by Foo Yun Chee and Sudip Kar-Gupta. This brief covers the Commission confirmation as reported by Reuters; it does not re-litigate the underlying Mythos or Astra eval disclosures.