OpenAI said Monday it would grant the European Union access to its new cybersecurity AI model, GPT-5.5-Cyber, while Anthropic has yet to offer the bloc similar access to its own model, Mythos.
European Commission spokesperson Thomas Regnier said the Commission welcomed OpenAI's engagement, including its intent to grant access to the new model. Despite having met with Anthropic on four or five occasions, the Commission had not yet entered into any conversations about obtaining access to that company's AI models, Regnier said.
"With one [OpenAI], you have a company proactively offering to give access to the company. With the other one [Anthropic], we have good exchanges though we're not at a stage where we can speculate on potential access or not," Regnier said at a press briefing, according to CNBC.
Additional meetings between OpenAI and the Commission are scheduled for this week, Regnier said. He quoted the Commission's position that model access would enable it to "follow deployment of the model very closely, and address security concerns." On Anthropic, Regnier acknowledged ongoing contact but said the two situations were not comparable, describing the Anthropic talks as being at a "different stage" and "not yet at the same stage as the solution we have on the table from OpenAI."
Access to GPT-5.5-Cyber will be extended to a range of European stakeholders, OpenAI said, including businesses, governments, cybersecurity authorities, and EU bodies such as the EU AI office. The model is being made available through a limited preview aimed at vetted cybersecurity teams.
In a statement, George Osborne, OpenAI's head of OpenAI for Countries, called for shared responsibility in cybersecurity, arguing that AI developers alone should not determine what is safe. An EU Cyber Action Plan, he said, would serve as the vehicle for broadening European access to defensive tools in partnership with policymakers, institutions, and businesses.
OpenAI's model arrived about a month after Anthropic announced Project Glasswing, a cybersecurity initiative built around Claude Mythos Preview, a model Anthropic says can find and exploit software vulnerabilities at a level matching top human security researchers. Anthropic restricted Mythos to 12 launch partners — including Amazon $AMZN Web Services, Apple $AAPL, Cisco $CSCO, CrowdStrike $CRWD, Google $GOOGL, Microsoft $MSFT, and Nvidia $NVDA — all of whom agreed to use it exclusively for defensive security work, with more than 40 additional organizations also receiving access.
Anthropic has said Mythos Preview identified thousands of previously unknown vulnerabilities across major operating systems and browsers during testing, including a flaw dating back 27 years in OpenBSD. The model's existence became public after internal draft documents were inadvertently left in an unsecured location on Anthropic's servers, a lapse the company attributed to a misconfiguration in a third-party content management tool.