Ukraine Seeks AI Sovereignty, Rejects Remote Control

Ukraine is strategically prioritizing the use of artificial intelligence systems that can be hosted on its own infrastructure, a move driven by wartime security concerns. This self-hosted, or 'on-premise,' approach ensures that critical government, business, and military digital tools remain under Ukrainian control, mitigating risks associated with remote systems that providers could restrict or deactivate.

The ministry's stance was reinforced by recent global events, including a U.S. government order impacting powerful AI models and potential curbs being considered in China. These developments underscore for Ukraine the critical need for 'AI sovereignty,' which is viewed not merely as a talking point but a practical necessity.

While Ukraine's current AI assistant in the Diia app utilizes Google's Gemini, accessed via EU servers, it's considered an interim solution. Personal data is stripped before queries to Gemini due to a lack of direct model control. Looking ahead, Ukraine is collaborating with Kyivstar to develop its own AI model based on Google's open-source Gemma variant, expected in autumn. This domestic model is intended for broad application across government services, private enterprises, and the military.