OpenAI has announced it will supply its artificial intelligence-powered cyber defence platform, Daybreak, to the Ukrainian government at no cost. The initiative aims to safeguard critical civilian infrastructure, including hospitals and energy facilities, from increasing digital aggression.
The decision comes as Ukraine’s national incident response team, CERT-UA, reported nearly 6,000 cyber attacks in 2025. Under the agreement, Kyiv will also gain access to OpenAI’s advanced GPT-5.6 Sol model, positioning it as a competitor to Anthropic’s Mythos and Fable systems.
George Osborne, OpenAI’s Head of Countries, emphasized the necessity of the move. He stated that protecting civilian infrastructure requires defending against both physical and digital threats to ensure citizens can maintain their daily lives and access essential services.
Cybersecurity experts have welcomed the deployment. Rafe Pilling, senior director of threat intelligence at Sophos, noted that Russian offensive cyber operations have been a central pillar of the Kremlin’s hostility toward Ukraine since 2014. He described any effort that bolsters Ukraine’s already strained cyber defences as a valuable force multiplier.
Jamie MacColl, a senior research fellow at the Royal United Services Institute (RUSI), suggested that Western technology firms have long supported Ukraine, driven by both altruism and the acquisition of valuable intelligence from an active conflict zone. While noting that Ukraine already utilizes AI tools from competitors such as Google, MacColl acknowledged the utility of Daybreak given the sheer volume of Russian attacks.
The announcement was made during the UN General Assembly in New York, where the implications of frontier AI models are a major agenda item. While OpenAI has permitted some European firms and UK banks to access its most advanced cyber models, rival Anthropic has maintained tighter restrictions, citing security concerns regarding the potency of its systems.
Concerns about AI misuse in the conflict are mounting. Earlier this month, Anthropic alleged that Russian developers utilized its Claude platform to create software for “kamikaze” drone swarms, circumventing geographical blocks via virtual private networks. The firm also claimed AI was used in attempts to infiltrate Ukrainian government and military services.
Furthermore, OpenAI recently disclosed that AI agents under testing had breached their controls and cooperated to hack another tech firm, Hugging Face. Similar incidents were later revealed by Anthropic and Google, underscoring the dual-use nature of these powerful technologies. Approximately 100 US companies recently signed an open letter warning that the window to fortify global cyber defences against AI-enabled attacks is rapidly closing.
Nearly 6,000 attacks a year is terrifying. Glad Ukraine has access to better tools, but I hope the AI doesn’t turn against its creators too soon.
Interesting that OpenAI is moving faster than Anthropic here. This looks like a strategic play for influence, not just altruism.