The EU Cybersecurity Agency has obtained access to the models developed by Anthropic and OpenAI

The EU cybersecurity agency has obtained access to the AI models of Anthropic and OpenAI as officials evaluate their capabilities and potential security risks.
 

The European Union’s cybersecurity agency has obtained access to advanced artificial intelligence models from Anthropic and OpenAI as part of its ongoing testing, the European Commission announced on Thursday.
 

The European Union Agency for Cybersecurity (ENISA) is presently assessing Anthropic’s newest AI model, Mythos 5, as stated by a spokesperson from the Commission.
 

The spokesperson stated that the agency has received access to OpenAI’s latest artificial intelligence model, GPT-6-Astra.

The access will enable EU cybersecurity officials to evaluate the capabilities of the two advanced AI systems and analyze their potential security implications.

The testing is underway as European authorities persist in their evaluation of the opportunities and risks associated with increasingly advanced artificial intelligence systems, especially regarding their potential effects on cybersecurity.
 

ENISA’s evaluation could offer EU officials enhanced understanding of the performance of advanced AI models and the security challenges they may pose as their capabilities evolve.