OpenAI has expanded access to its latest AI models — including GPT-5.5-Cyber to major European companies as part of a push to strengthen cyber resilience across critical industries.
The programme includes organisations such as Deutsche Telekom, BBVA, Telefónica, Sophos and Scalable Capital.
The initiative forms part of OpenAI’s “Trusted Access for Cyber” programme, which provides verified organisations in sectors such as finance, telecoms, energy and public services with controlled access to advanced AI tools designed for defensive cybersecurity work.
According to Emmanuel Marill, OpenAI’s managing director for EMEA, the company is attempting to balance usefulness with safety as AI systems become more capable. He said trusted defenders need effective tools to identify vulnerabilities and respond to threats quickly, while dangerous uses must be restricted.
The move comes amid growing concern over the cybersecurity implications of frontier AI systems. Last month, rival Anthropic released its Mythos model, which experts say significantly increases both defensive and offensive cyber capabilities because of its advanced coding and vulnerability discovery skills.
European regulators are also taking a closer interest. OpenAI has reportedly offered the European Commission open access to some cybersecurity features, while officials suggested Anthropic has been less open in its cooperation.
Former UK chancellor George Osborne, who now leads OpenAI’s “OpenAI for Countries” initiative, said broader access to defensive AI tools could strengthen shared security and align with European priorities around public safety and digital resilience.
