The start date of NATO's use of the models is not public, and the usage figures and effectiveness claims come only from OpenAI and have not been independently verified. NATO has not published its own account of the deployment. NCIA general manager Dylan Browne has praised the models in public remarks, calling them a major step forward, but no NATO document describing the scope, contract terms, or security controls has been released. Earlier reporting said OpenAI sought to place its technology on NATO's unclassified networks and that Altman had told staff the company wanted to deploy across all NATO classified networks. NATO did not respond to a request for comment on that report. The Cooperative Cyber Defence Centre of Excellence (CCDCOE) is also involved: OpenAI supported its 2026 Crossed Swords exercise, which included a training stream using Daybreak in which participants built and evaluated AI agents for cyber operations.
This is a large, public step by a commercial AI lab into a military alliance's cyber operations, and it comes as European governments weigh how much frontier AI to rely on for defense. Attorneys advising clients on government contracts, data handling, or AI procurement should watch for three things: whether NATO or NCIA publishes contract terms or a formal authorization; whether the classified-network expansion discussed in earlier reporting moves forward; and how the arrangement interacts with OpenAI's parallel work with the Pentagon on a classified network. Commentators have also raised questions about Europe's dependence on American technology providers, a theme likely to surface in procurement and data sovereignty debates. OpenAI has also given its cyber models to the EU's cybersecurity agency ENISA and to national cyber agencies in France, Germany, and Poland, which expands the footprint beyond NATO itself.