EU Cybersecurity Agency Granted Access to Anthropic's Mythos 5 and OpenAI's GPT-6 Astra
A European Commission spokesperson confirmed that the European Union Agency for Cybersecurity (ENISA) has been granted testing access to high-capability AI models from Anthropic and OpenAI. ENISA is currently evaluating Anthropic's Mythos 5 as well as OpenAI's latest model, GPT-6 Astra. This access allows European regulators to directly evaluate state-of-the-art AI capabilities and potential security risks, particularly around critical software defense. It also marks a significant step in cross-border AI governance following prior U.S. restrictions on foreign access to high-performance models. The testing access comes after ongoing debate over U.S. White House policy regarding foreign organizational access to powerful domestic AI technologies. Both Mythos 5 and GPT-6 Astra exhibit state-of-the-art capabilities in automated coding, vulnerability discovery, and cybersecurity operations.
## BACKGROUND
Frontier AI models like Anthropic's Mythos series and OpenAI's GPT-6 Astra possess enhanced reasoning capabilities that can be utilized for finding software vulnerabilities and cyber defense. Due to concerns over dual-use safety risks, deployment of these restricted models is typically limited to vetted security evaluation programs and government partners.