OpenAI in talks with EU over access to cyber model; Anthropic still holding out on Mythos

The European Commission said it was in talks with the European Union about access to OpenAI’s new cyber model, but Anthropic was still resisting publishing Mythos to the block.

OpenAI announced Thursday that a variant of its latest AI model, GPT-5.5-Cyber, is available to vetted cybersecurity teams in limited preview capacity. It comes a month after Anthropic launched its own model, Mythos, which sparked a wave of fear about cyberattacks on critical software.

“We welcome OpenAI’s transparency and its intention to provide the commission with access to the new model,” Commission Spokesperson Thomas Regnier said at a press conference on Monday. he said.

He confirmed that an exchange of ideas had taken place between OpenAI and the EU and that further talks on access to the model were planned this week.

“This will allow us to track the deployment of the model very closely and address any security concerns,” he added.

Even though Mythos launched a month ago, Anthropic has yet to grant preview access for AB to review.

Regnier added that the EU had discussed access with Anthropic, but the talks were “at a different stage” than with OpenAI.

OpenAI and Anthropic were approached for comment.

This is a story of breaking. Refresh for updates.