Oossa

OpenAI gives NATO and European governments access to Daybreak cybersecurity tool

OpenAI lets NATO, ENISA, France, Germany, Poland and Ukraine use its AI‑driven Daybreak system to hunt software bugs and patch gaps.

By Published by Oossa: 2 min read

Compagnons · Unsplash

OpenAI announced that its Daybreak cybersecurity platform is now available to NATO and a handful of European governments. The move lets national cyber‑defense teams run OpenAI’s large language models – the kind of AI behind ChatGPT – on their own code bases. The goal is to speed up audits, surface hidden vulnerabilities and suggest fixes before attackers can exploit them. NATO’s Communications and Information Agency (NCIA) says it is already using the underlying models heavily, and its chief Dylan Browne called the deployment a generational leap for cyber defense. The first public test of the system came during NATO’s Crossed Swords 2026 exercise, where participants deployed AI agents for cyber operations under human oversight for the first time. The same tools have been used by the EU’s cybersecurity agency ENISA, which reported four bugs in software that runs across EU institutions. A joint analysis with CERT‑EU covered 27 software projects; of the 88 issues flagged, 95 % turned out to be real bugs. Poland’s CERT Polska uncovered six flaws in MikroTik routers that were already being exploited in the wild. Ukraine is also using Daybreak to protect its civilian infrastructure, and France, Germany and Poland have been granted access alongside NATO. OpenAI has not disclosed pricing or licensing details for the government rollout. The company says the tool runs on its own secure cloud infrastructure, and access is limited to vetted national cyber teams. No independent third‑party evaluation of Daybreak’s effectiveness has been published yet, so the broader security community is waiting for more data. The partnership reflects a growing trend of governments turning to commercial AI firms for advanced cyber‑tools. It also shows OpenAI’s willingness to expand beyond the consumer market into critical national‑security applications.

What Daybreak does and why it matters

Daybreak combines code‑analysis models with a large language model that can understand software documentation, trace data flows and suggest remediation steps. In practice, a cyber analyst can feed a piece of code into the system and receive a list of potential weaknesses, ranked by severity, along with suggested patches. The system can also scan large code repositories automatically, flagging patterns that often lead to security bugs. For ordinary citizens, the rollout could mean more resilient public services. If a government can patch a vulnerability in a health‑record system or a utility‑grid controller before a hacker finds it, the risk of data leaks or service outages drops. However, the effectiveness of the tool in real‑world attacks remains unproven until independent security researchers test it. OpenAI’s move also raises questions about data privacy and the handling of sensitive code. The company says it runs the models in a closed environment and does not retain customer data, but the exact safeguards have not been detailed publicly.

Why it matters

For citizens, faster bug detection could make public digital services harder for hackers to breach, lowering the chance of data loss or service disruption. The rollout also shows governments relying on commercial AI for security, which may speed up defenses but also brings questions about data handling and independent verification.

Was this article useful?
Share

Read next

Oossa · Newsletter

The week in AI, explained

Every Monday: the stories worth knowing, in plain language. Free, no spam.