OpenAI Gives NATO and Allied Cyber Agencies Access to Advanced Cyber Models
OpenAI described its work with NATO and allied cyber agencies to help trusted defenders find vulnerabilities and test fixes.
- NATO: through the NATO Communications and Information Agency (NCIA), NATO’s core institutions have access to advanced cybersecurity models; NATO has used up to 300 billion raw input and output tokens per month.
- Exercise: OpenAI supported the 2026 Crossed Swords exercise run by the NATO CCDCOE, adding a training stream where participants built, deployed and evaluated AI agents for cyber operations using OpenAI Daybreak.
- Other defenders with access: the EU’s ENISA and national cyber agencies in France (ANSSI), Germany and Poland. Ukraine’s Ministry of Digital Transformation uses Daybreak for civilian infrastructure defense.
- ENISA used the models to identify four vulnerabilities in software used across EU institutions; fixes were published for all four.
- A November ENISA and CERT-EU report reviewed 27 software projects with frontier models (including OpenAI models via Daybreak and the Codex security plug-in). Of 88 potential issues assessed by maintainers, 95% were confirmed genuine.
- CERT Polska used OpenAI models to help find six vulnerabilities in MikroTik RouterOS; MikroTik released fixes, which CERT Polska confirmed stopped the attacks it had observed.