The initiative enables companies to build and test autonomous drones and decision-support systems on operational battlefield datasets.
Ukraine is opening partner access to train AI models using real battlefield data — the first initiative of its kind in the world.
According to Mykhailo Fedorov, the Minister of Defence of Ukraine, the government has approved a resolution launching a new framework for cooperation between the state, Ukrainian companies, and international partners.
For startups, the initiative opens the opportunity to develop and validate defence AI systems using real-world operational data rather than simulated environments.
Early-stage companies working on autonomous drones, computer vision, electronic warfare resilience, and battlefield decision-support tools will be able to train and refine algorithms on large-scale, continuously updated datasets generated during active operations.
Access to this type of data — typically restricted or unavailable outside military programmes — could significantly shorten development cycles and improve model performance in real-world conditions.
It also positions Ukraine as a unique, if involuntary, testbed for defence-tech innovation, creating potential pathways for startups to collaborate directly with government agencies, integrate their technologies into operational systems, and accelerate the commercialisation of next-generation autonomous defence platforms.
Fedorov asserts:
“In modern warfare, we must outperform Russia in every technological cycle. AI is one of the key arenas of this competition.
The future of warfare belongs to autonomous systems. Our objective is to increase the level of autonomy in drones and other combat platforms so they can detect targets faster, analyse battlefield conditions, and support real-time decision-making.”
High-quality data for training neural networks is critical, and the Ministry of Defence has established a dedicated AI platform at the Centre for Innovation and Development of Defence Technologies.
The platform enables partners to: Securely train models without direct access to sensitive databases; Work with large volumes of labelled photo and video data; Use datasets that are continuously updated.
Ukraine currently possesses a unique body of battlefield data unmatched anywhere in the world. This includes millions of annotated frames collected during tens of thousands of combat drone missions. These datasets are already used to train neural networks that automatically detect ground and aerial targets within the DELTA system.
International partners and Ukrainian companies have expressed strong demand for precisely this type of data to develop and modernise defence technologies.
For Ukraine, this initiative represents the next stage of a win-win partnership model. Partners gain the opportunity to train their AI models on real data from modern warfare, while Ukraine accelerates the development of autonomous systems and delivers new technological capabilities to the front line.
Fedorov asserts:
“We are ready to work with partners on joint analytics, model training, and the development of new technological solutions.”
Lead image: Freepik