OpenAI details AI support for NATO and allied cyber defense
OpenAI says it works with NATO’s Communications and Information Agency to provide NATO’s core institutions access to advanced cybersecurity models for protecting Allied networks. The company also describes Daybreak, its program for qualified defenders conducting authorized security work.

Key takeaways · 4
- 01
OpenAI says NATO’s core institutions can access its advanced cybersecurity models through work with NATO’s Communications and Information Agency.
- 02
NATO’s work has used up to 300 billion raw input and output tokens per month, according to OpenAI.
- 03
Daybreak is for qualified enterprise customers and cybersecurity practitioners doing authorized cybersecurity work; safeguards and access controls continue to apply.
- 04
OpenAI says ENISA used its models to identify four software vulnerabilities, and fixes were published for all four.
NATO access and scale
OpenAI says it works with NATO’s Communications and Information Agency to give NATO’s core institutions access to advanced cybersecurity models for protecting Allied networks.[1] The work has used up to 300 billion raw input and output tokens per month, according to OpenAI.[1] OpenAI says its models are also available to European defenders, including the EU’s ENISA and national cyber agencies in France, Germany and Poland.[1]
Daybreak’s authorized-use boundaries
OpenAI describes Daybreak as a program that gives trusted defenders advanced AI for authorized security tasks, including software review, investigating suspicious activity, validating vulnerabilities and testing fixes.[1] The program is for qualified enterprise customers and cybersecurity practitioners doing authorized cybersecurity work.[2] OpenAI says users should work only on systems, applications, accounts, networks or data they own, operate or are explicitly authorized to test or analyze.[2] Its usage policies, safeguards and access controls continue to apply.[2]
AI in a cyber defense exercise
The NATO Cooperative Cyber Defence Centre of Excellence said Crossed Swords 2026 ran from November 2 to 6 and brought together more than 250 participants from over 40 NATO and non-NATO member states.[3] Participants designed, developed, deployed and evaluated AI agents, working alongside them as operators.[3] The centre said the exercise used on-premises AI and OpenAI Daybreak for the first time.[3] Its AI training emphasized maintaining human oversight in complex, multi-domain missions.[3]
Reported vulnerability findings
OpenAI says ENISA used its models to identify four vulnerabilities in software used across EU institutions, and fixes were published for all four.[1] OpenAI also says ENISA and CERT-EU’s November joint report described a review of 27 software projects using frontier models, including OpenAI models accessed through Daybreak.[1] Software maintainers confirmed 95% of the 88 potential issues they assessed across the tested models as genuine, according to OpenAI.[1] Separately, OpenAI says CERT Polska helped identify six vulnerabilities in MikroTik RouterOS and coordinated disclosure with the vendor.[1]
For security leaders, the announcement points to AI-assisted vulnerability work being used alongside human operators, not as a replacement for their oversight. Teams considering similar tools should distinguish demonstrated findings from broader claims and confirm authorization, safeguards and validation practices before use.
Why it matters
AI training for your team
kju.ai is built for teams: daily sessions, team analytics, and rollout support for your organization.
How this developed
17 November 2026
OpenAI details AI support for NATO and allied cyber defense
Sources
- Strengthening NATO and allied cyber defenses with AI | OpenAIopenai.com
- OpenAI Daybreak - Trusted Access for Cyber overview | OpenAI Help Centerhelp.openai.com
- CCDCOEccdcoe.org