OpenAI is reportedly in talks with the European Union over access to its new cyber-focused model, while Anthropic has yet to release its Mythos model to the bloc, according to the European Commission.

On Thursday, OpenAI said GPT-5.5-Cyber, a specialised version of its latest AI system, is being made available in a limited preview to selected cybersecurity teams.

This follows Anthropic’s release of Mythos last month, which sparked concerns about its potential use in cyberattacks targeting critical software systems.

“We welcome OpenAI’s transparency and intent to give commission access to new model,” Commission Spokesperson Thomas Regnier said at a press briefing on Monday.

He confirmed that an exchange had already taken place between OpenAI and the European Union, with additional discussions scheduled for this week regarding access to the model.

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

Although Mythos was released a month ago, Anthropic has yet to grant the European Union preview access to review the model.

The EU is currently in discussions with Anthropic over access, Regnier said, but noted that those talks are at a “different stage” compared with its engagement with OpenAI.