Ukraine Developing AI System with Google Open-Source Tools – Ministry
Ukraine is developing a large language model using Google’s open-source Gemma framework to create an independent AI system for both military and civilian applications, the country’s digital ministry and mobile operator Kyivstar said on Monday.
The project will initially use Google’s computing infrastructure for training before moving entirely to local servers, ensuring Ukraine retains full control over AI systems accessed by 23 million citizens daily. The initiative also addresses gaps in current AI tools, such as handling local dialects combining Ukrainian, Russian, and Bulgarian, and correcting mistranslations in legal documents.
Four advisory committees have been set up to oversee technical, legal, cultural, historical, and linguistic aspects, ensuring the model supports Ukrainian, minority languages like Crimean Tatar, and Russian. Data is being collected from over 90 government institutions, including court registries, educational publishers, regional archives, and records of Russian actions in the ongoing war.
Training will take place on secure GPUs outside Ukraine provided by Google before deployment in local data centers. Kyivstar has not confirmed a launch date but noted that the system will initially support government and company platforms, with eventual expansion to the private sector.
Security is a key concern, as Russian cyberattacks continue. Measures against threats such as “prompt injection” attacks are being developed. Kyivstar, the first Ukrainian company to list on a U.S. stock exchange, has also installed over 3,500 backup generators to stabilize operations amid strikes on the energy grid.
Officials say the project demonstrates how smaller nations can leverage open technologies to achieve strategic independence in AI, reducing reliance on foreign systems.
