Ukraine is the first country in the world to open real battlefield data to partners for AI model training

The Cabinet of Ministers of Ukraine has adopted a resolution that launches a new format of cooperation between the government, Ukrainian companies, and international partners. Ukraine is the first country in the world to open the possibility of training AI models for unmanned systems using real battlefield data. 

“In modern warfare, we must defeat russia in every technological cycle. Artificial intelligence is one of the key areas of this competition.

The future of warfare belongs to autonomous systems. Our task is to enhance the autonomy of drones and other combat systems so that they can detect targets more quickly, analyze the situation, and help inform decisions on the battlefield,” stressed Ukraine’s Minister of Defence, Mykhailo Fedorov. 

According to him, the key to this is high-quality data for training neural networks. That is why earlier this year, a dedicated AI platform was set up at the MoD’s Center for Innovation and Development of Defense Technologies, enabling the following:

• secure training of models without direct access to sensitive databases;
• working with large volumes of annotated photo and video materials;
• using data that is continuously updated. 

Ukraine today possesses a unique battlefield dataset unavailable anywhere else in the world. It includes millions of annotated frames collected during tens of thousands of combat flights. These data are already used to train neural networks that automatically identify ground and aerial targets in the DELTA system. 

International partners and Ukrainian companies have a strong demand for this kind of data to develop and modernize defense technologies. 

“For us, this is the next step in developing a win-win partnership. Partners gain the opportunity to train their AI models on real data from modern warfare. At the same time, Ukraine gains faster progress in the development of autonomous systems and new technological solutions for the front.