The European Commission is in active talks with two of the world’s most powerful AI companies over access to their most advanced models, and the conversations are going very differently depending on which company you ask.

OpenAI has committed to granting EU cybersecurity defenders access to its GPT-5.5-Cyber model, a move that Brussels has received warmly. Anthropic, meanwhile, has been in four to five meetings with Commission officials, but those discussions haven’t yet produced similar access to its Mythos system.

OpenAI takes the cooperative lane

That’s essentially what OpenAI has done. The company’s GPT-5.5-Cyber model is being positioned as a cybersecurity asset for European institutions and defenders. EU officials have expressed strong support for the offer, which would give government-aligned security teams access to advanced AI capabilities designed to identify and respond to threats.

Anthropic’s slower road

Anthropic’s situation tells a different story. Four to five meetings with Commission officials is not nothing, but it hasn’t translated into the kind of concrete access that OpenAI is offering.

The Mythos system, Anthropic’s advanced model at the center of these discussions, remains out of reach for EU defenders for now.

Why this matters beyond Brussels

The Commission’s engagement with these AI firms is about more than cybersecurity tooling. It’s a signal about how Europe intends to integrate frontier AI models into its institutional infrastructure.

Europe has positioned itself as the global leader in AI regulation through the AI Act, with enforcement phases starting in 2025. By negotiating direct access to models from OpenAI and Anthropic, the Commission is acknowledging that it needs the private sector’s most advanced technology to defend against increasingly sophisticated cyber threats.

This dynamic has implications well beyond government cybersecurity. The same AI models being discussed for defending European institutions could eventually play roles in securing financial infrastructure, including the digital asset ecosystem. Crypto platforms, DeFi protocols, and blockchain networks are frequent targets for sophisticated attacks, though no direct connections to these AI tools have been established yet.