OpenAI gives cyber defence tools to Ukraine

OpenAI says it will provide its Daybreak cyber defence system to the Ukraine government at no cost, a move aimed at protecting civilian infrastructure such as hospitals and power plants from cyber-attacks. The company is also giving Ukraine access to GPT 5.6 Sol, its advanced model, under the same deal.

The timing is blunt. Ukraine’s CERT-UA, the national cyber incident response team, says the country recorded nearly 6,000 attacks in 2025. Daybreak is designed to spot weaknesses in digital systems quickly and help produce fixes, which sounds less glamorous than AI chatbots and a lot more useful when a power grid is under pressure.

George Osborne, now OpenAI’s Head of OpenAI for Countries, framed the deal as part of defending civilian life itself. That fits the shape of this war, where cyber operations have been part of the fight for years. Rafe Pilling of Sophos said Russian offensive cyber operations have been a key part of the Kremlin’s aggression since 2014, and that bolstering Ukraine’s cyber defence would be a welcome force multiplier.

Jamie MacColl of RUSI said western tech companies have been supporting Ukraine for some time, for both altruistic and self-interested reasons, since active conflicts generate valuable data and intelligence. He said Daybreak should be useful given the scale of Russian attacks, though Ukraine already uses AI tools from other firms, including Google. The message is simple: in this war, the software stack matters.

The bigger issue is that the same frontier models that help defenders can also help attackers. Anthropic has tightened access to its systems because of that risk, while OpenAI has been more open with some firms in Europe and UK banks. And as the UN General Assembly gets underway in New York, AI is getting treated less like a lab toy and more like a geopolitical tool with sharp edges.