March 17, 2026

Ukraine is opening access to real battlefield data to train artificial intelligence models used in unmanned systems
The initiative relies on a unique dataset collected during the full-scale war, which could accelerate the development of autonomous technologies for modern battlefields.
This was announced by Minister of Defence of Ukraine Mykhailo Fedorov in a Facebook post.
According to him, the government has adopted a resolution launching a new format of cooperation between the state, Ukrainian companies, and international partners. For the first time in the world, foreign allies will be able to train AI models for unmanned systems using real combat data.
“Ukraine is opening the possibility to train artificial intelligence models for unmanned systems using real battlefield data,” Fedorov emphasized.

The main goal of the initiative is to increase the autonomy of drones and other combat systems so they can detect targets faster, analyze the situation, and assist in decision-making directly on the battlefield.
According to the minister, the future of warfare will largely depend on autonomous systems. The key to developing them lies in high-quality datasets used to train neural networks.
To support this effort, a dedicated AI platform has been created at the MoD’s Center for Innovation and Development of Defense Technologies. The platform allows partners to:
- safely train AI models without direct access to sensitive databases;
- work with large volumes of annotated photo and video materials;
- use datasets that are continuously updated.
Ukraine possesses a unique body of battlefield data that does not exist anywhere else in the world. It includes millions of annotated frames collected during tens of thousands of combat drone flights.
These datasets are already being used to train neural networks capable of automatically detecting ground and aerial targets within the DELTA battlefield management system.

Foreign partners and defense companies have long shown strong interest in such data. Large and diverse datasets are essential for training algorithms that recognize patterns, objects, and battlefield behavior of both people and equipment.
Ukraine views this cooperation as mutually beneficial. Partners gain the opportunity to train their AI models using real data from a high-intensity modern war, while Ukraine benefits from faster development of autonomous systems and new technological solutions for the battlefield.
According to Fedorov, Ukraine is ready to work with partners on joint analytics, AI model training, and the development of new technologies for unmanned systems.
