█

AI/TLDR

OpenAI · 2026-11-17 · notable

OpenAI Daybreak for NATO — allied cyber agencies get OpenAI's defense models

In short

OpenAI says NATO's core institutions now use its Daybreak cyber defense models through the NATO Communications and Information Agency, at up to 300 billion tokens a month. ENISA and agencies in France, Germany and Poland also have access.

OpenAI · 2026-11-17 · 2 sources checked

StartupHub.ai article card for OpenAI putting Daybreak to work for NATO

NATO, ENISA and three national cyber agencies use OpenAI's Daybreak models to find and patch software flaws.

What is it?

OpenAI's Daybreak cyber defense models are now in use at NATO's core institutions, with access running through the NATO Communications and Information Agency (NCIA). OpenAI also names the EU cyber agency ENISA and national agencies in France, Germany and Poland as users, and says it supported NATO's 2026 Crossed Swords exercise.

How does it work?

Daybreak is OpenAI's defender toolkit: an analyst submits suspect code, the models flag possible flaws, the analyst checks them on real systems and then tests a patch. At Crossed Swords, run by NATO's Cooperative Cyber Defence Centre of Excellence, teams built AI agents with Daybreak while humans stayed the operators.

Why does it matter?

The early results are concrete: ENISA found four vulnerabilities in software used across EU institutions, all now fixed, and CERT Polska helped find six flaws in MikroTik RouterOS. In a joint ENISA and CERT-EU review of 27 projects, maintainers confirmed 95% of 88 flagged issues, so human checks still decide what is real.

Who is it for?

government security teams and vulnerability researchers

Sources · 2 outlets

Tags

  • openai
  • daybreak
  • nato
  • ncia
  • enisa
  • cert-polska
  • cybersecurity
  • vulnerability-research
  • government
  • crossed-swords

← All releases · Learn AI