OpenAI on November 17, 2026 published an account of how it is putting cybersecurity models into the hands of NATO and allied government defenders through Daybreak, the company's program for authorized security work. The post is a deployment and partnership update, not a new model launch. OpenAI already introduced GPT-6 Cyber for Daybreak Red on November 5.

The company said it works with NATO's Communications and Information Agency (NCIA) to give NATO's core institutions access to advanced cybersecurity models, and that NATO has used up to 300 billion raw input and output tokens per month as part of this work. OpenAI framed the effort as help for multinational teams that protect Allied networks around the clock, and as a source of lessons for defenders across Europe and the wider Alliance.

OpenAI also said it supported the 2026 Crossed Swords exercise organized by the NATO Cooperative Cyber Defence Centre of Excellence (CCDCOE). According to the company, the exercise included a new training stream in which participants built, deployed and evaluated AI agents for cyber operations, using Daybreak in the technical environment, with an emphasis on keeping human oversight. NCIA general manager Dr Dylan Browne was quoted as saying that AI-powered cyber defence models accelerate vulnerability discovery and threat detection, and that NATO will study token consumption, usage patterns and the economics of the models as it scales adoption. Sasha Baker, OpenAI's head of national security policy, said the company wants to put more capable tools in the hands of NATO's cyber defenders and to deepen the work with the Alliance and its allies.

Beyond NATO, OpenAI said it has already provided access to its cyber models to defenders in Europe, including the European Union Agency for Cybersecurity (ENISA) and national cyber agencies in France (ANSSI), Germany and Poland. It said Ukraine's Ministry of Digital Transformation is using Daybreak to help identify software vulnerabilities and develop and test fixes for civilian infrastructure under persistent cyber attacks.

OpenAI attributed several concrete findings to those partners. It said ENISA used the models to identify four vulnerabilities in software used across EU institutions, with fixes published for all four. It also pointed to a November joint report from ENISA and CERT-EU that, as OpenAI described it, reviewed 27 software projects using frontier models including OpenAI models accessed through Daybreak, and tools including the Codex security plug-in; across the models tested, 95 percent of the 88 potential issues that software maintainers assessed were confirmed as genuine, according to OpenAI's summary of that report. This run did not open the underlying ENISA and CERT-EU documents. OpenAI further said CERT Polska used its models to help identify six vulnerabilities in MikroTik RouterOS, coordinated disclosure with the vendor, and confirmed that MikroTik's fixes prevented the attacks CERT Polska had observed.

The figures, partner results and quotations all come from OpenAI's own page. The company said it remains committed to NATO's principles for responsible AI adoption.

References