OpenAI Seals EU Cyber Deal as Anthropic Holds Back
The European Union wanted access to both. It only got one. On May 11, 2026, OpenAI agreed to grant the EU access to its new cyber AI model. Anthropic, however, continues to hold back Mythos from European regulators and institutions.
OpenAI EU Cyber Model: What the Deal Covers
OpenAI will extend GPT-5.5-Cyber access to European businesses, governments, cyber authorities, and EU institutions, including the EU AI Office. The model rolled out last week in a limited preview to vetted cybersecurity teams. George Osborne, OpenAI’s Head of OpenAI for Countries, explained the rationale. “AI labs like ours shouldn’t be the sole arbiters of cyber safety,” he said. He added that defensive AI tools should reach Europe’s many defenders, not just a privileged few.
Where Anthropic Stands on Mythos
European Commission spokesperson Regnier confirmed the EU discussed access with Anthropic. However, he noted talks are at a “different stage.” The Commission held four or five meetings with Anthropic. Despite that, progress lags well behind the OpenAI deal. Anthropic released Mythos last month through Project Glasswing, limiting access to approximately 40 organisations, including Apple, Amazon, JPMorgan Chase, and Palo Alto Networks. The company considers a full public release too risky because of the scale of cyberattacks the model could enable.
UK AI Security Institute tests placed Mythos slightly ahead of GPT-5.5-Cyber. Mythos completed a 32-step simulated corporate cyberattack in 3 out of 10 runs. GPT-5.5-Cyber succeeded in 2 out of 10. Before Mythos launched, no AI model had ever passed that test. Both companies have now restricted access to their cyber models, despite OpenAI CEO Sam Altman previously criticising Anthropic for doing exactly that. Meanwhile, JPMorgan’s Jamie Dimon warned that AI tools are first making companies more vulnerable before they can defend themselves. That tension sits at the heart of the EU’s negotiations with both labs.
