Artificial Intelligence

OpenAI puts Daybreak to work for NATO

OpenAI gave NATO cyber defenders Daybreak models that found real flaws, with NATO using up to 300 billion tokens a month.

StartupHub.ai Staff

By StartupHub.ai Staff, reviewed by Daniel Singer, Editor

2 min read

NATO cyber defenders using AI models for network defence
Credit: OpenAI News

OpenAI gave NATO core institutions access to cyber defence models through Daybreak for authorized work on Allied networks.

Defenders protect hospitals, power systems and ministries that break quickly when hidden software flaws stay unpatched.

Daybreak is OpenAI defender toolkit that reviews code, investigates suspicious activity and tests fixes before use, OpenAI said.

An analyst feeds suspect code to the model, checks flagged flaws on real systems and then develops a patch, OpenAI said.

NATO has used up to 300 billion tokens, or billed units of model input and output, per month as of November 2026, OpenAI said.

Dr Dylan Browne, NCIA general manager who buys technology for the alliance, said the models find flaws at machine speed and aid resilience, according to OpenAI.

NATO Cooperative Cyber Defence Centre of Excellence added AI agents built with Daybreak to its Crossed Swords exercise with humans kept as operators, OpenAI said.

Beyond NATO, defenders at EU agency ENISA and national agencies in France, Germany and Poland already have access, OpenAI said.

ENISA used the models to find four vulnerabilities, or hidden software flaws attackers can exploit, in software used across EU institutions and fixes shipped for all four, OpenAI said.

A joint ENISA and CERT-EU review of 27 software projects confirmed 95% of 88 potential issues as genuine after maintainer checks, OpenAI said.

Poland CERT Polska used the models to help find six vulnerabilities in MikroTik RouterOS and verify fixes stopped observed attacks, OpenAI said.

Ukraine Ministry of Digital Transformation uses Daybreak to find civilian infrastructure flaws and test fixes faster amid persistent attacks, OpenAI said.

Los Alamos National Laboratory, which said year-two work with OpenAI will enhance Venado-GPT for lab missions.

Sasha Baker, OpenAI head of national security policy who speaks for its government work, said defenders should hold the most advanced tools to find flaws first, according to OpenAI.

Browne cautioned NATO still must learn token consumption, usage patterns and model economics to control costs and compare vendors, according to OpenAI.

Human checks still decide, since even strong results left about five percent of flagged issues unconfirmed by maintainers.

Watch NATO procurement and patch times to see if defender access turns token volume into faster fixes.

Mentioned in this story

Sources

Get the free daily brief every morning

Every morning the day's AI news, new AI startups and new funding rounds. Free and takes a few minutes to read, unsubscribe at any time.

Join 64,000 other subscribers.

TopicsOpenAINATONCIADaybreakENISACERT PolskaMikroTikDylan Browne

StartupHub.ai StaffDaniel SingerWritten by StartupHub.ai Staff, Editorial teamReviewed by Daniel Singer, EditorAbout

StartupHub.ai Staff, Editorial team

The staff writers of StartupHub.ai, ranging from investment analysts to avid AI tool users, early adopters and critical enthusiasts. Backgrounds span engineering, business and the arts. We hold every piece to rigorous standards of research and review.

Daniel Singer, Editor

Daniel Singer is an AI expert who has been writing about, covering and building AI technology since 2018.

More from Daniel Singer

Test your Agent Readiness now

Paste your website. How well ChatGPT, Claude and the other agents can read and use a site, in about a minute.

Watch

The videos behind the news, one flip at a time

The feed mixes the videos from our articles with startup demos and you flip through them like TV channels for free with no signup.