After four years effectively as an all-in-one laboratory, training ground and live arena for technology to fight its own war, Ukraine is taking page from the tech world and productising, with the launch of an artificial intelligence development environment for allies. On March 12, Kyiv turned its front line into a state-regulated engine for autonomous systems by opening partner access to train AI models on real battlefield data.
The move marks a strategic pivot for Kyiv. By formalising access to its battlefield data, the government is shifting from a passive subject of research to a sovereign platform owner.
The project – which is so new, it does not appear to have a name at this point – is based around a secure pipeline of annotated footage from the country’s combat missions. Companies from both Ukraine and allied nations in turn can use that footage to refine their technologies using the reality of modern combat.
It is not clear how Ukraine will charge for this access, although the project is emerging at a time when the country continues to need funds both to support its economy and to fight against Russia. The latter has seen Ukraine receive hundreds of billions of dollars in aid and loans from the US and Europe, according to research from the Council on Foreign Relations. However, the same research notes that the US has not legislated any new aid packages since the end of 2024, with the election of Donald Trump as President.
Housed within the Ministry of Defence’s Centre for Innovation, the project aims to bridge the “sim-to-real” gap — the chasm between a clean computer simulation and a messy, unpredictable battlefield. The difference can lead sophisticated Western drones to fail when confronted with unpredictable local conditions and Electronic Warfare (EW).
They can now draw on an enormous pool of real warfare information. Last year alone, Ukrainian drones recorded around 820,000 verified strikes against Russian targets. In roughly 240,000 of them, enemy personnel were killed or critically wounded. Meanwhile, the country’s Avengers AI platform detects upwards of 12,000 enemy targets every week. Developers can now access these sources and the data that they gather to train their systems on the movements of a real Russian turtle tank or a camouflaged Lancet launcher.
“Ukraine currently possesses a unique body of battlefield data unmatched anywhere in the world,” recently appointed Defence Minister Mykhailo Fedorov said in a statement. “This includes millions of annotated frames collected during tens of thousands of combat drone missions.”
The strategic data shift
To be clear, international defence tech companies have been working with Ukrainian data going back to when Russia’s full-scale invasion began in February 2022. One of the most visible players is US software firm Palantir, which recently partnered with the Ukrainian Ministry of Defence to develop the Brave1 Dataroom, a platform for testing and training AI models for the battlefield.
The new initiative marks a significant shift in such relationships. It moves Ukraine from providing single-vendor access to historical data to a multi-vendor platform for training on continuously updated, near-real-time data. According to Fedorov, the same datasets already train neural networks that automatically detect ground and aerial targets within DELTA, Ukraine’s vaunted battlefield management system.
The data’s value for training AI systems lies in its breadth and grit. In the four years since Russia’s full-scale invasion began, Ukraine has become the only place on Earth where Western tech is pitted against a peer adversary in a high-intensity EW environment.
Samuel Bendett, a Russian military expert at the CNA research organisation, recently highlighted the benefits of this experience. He estimated that by late 2024, Ukraine had already aggregated roughly 2 million hours of battlefield video — equivalent to about 228 years. That number has likely grown significantly since then. (For a point of comparison, training to build functional autonomous driving systems starts at millions of hours, according to NVIDIA researchers. In other words, Ukraine holds a critical amount of battlefield data for training battlefield autonomous systems.)
“Imagine training your drones to navigate a highly complex battlespace like the Ukrainian war zone,” Bendett said. “You would want as varied a dataset as possible, and nothing is more varied than hundreds of years’ worth of data from the most active battlefield in the world today.”
The “made in Ukraine” badge
The move builds on a long-term strategy Fedorov shared with Resilience Media in an interview last September. “It’s important for us to test in Ukraine, and for our partners to fully understand the innovation cycle that we are following,” he said. “Because everything that you see right now on the battlefield is made in Ukraine.”
Indeed, that “made in Ukraine” expertise is now a major export. In recent weeks, Kyiv have sent anti-drone specialists to the Middle East to assist in countering Iranian-made Shahed loitering munitions, which have been increasingly used in regional conflicts. By opening up its data, Kyiv is positioning itself as a central intelligence hub for global conflict.
However, the rapid commodification of war data also brings significant risks. Professor Mariarosaria Taddeo of the University of Oxford has warned that AI is a “fragile technology” and mishandling the chain of responsibility for autonomous strikes could lead to “catastrophic” repercussions.
Margarita Konaev, of Georgetown’s Centre for Security and Emerging Technology, has added that as targeting decisions move closer to machine control, the risk of “inadvertent escalation” or errors increases. Opening millions of frames of combat footage to private firms also raises concerns about unsafe data handling.
To mitigate the risks, the Ministry will restrict direct access to sensitive databases, it said. Training will then take place within a secure, government-controlled environment.
In exchange for allowing allies into the vault, Ukraine expects to accelerate the development of its own autonomous systems and to deliver new technological capabilities to the front line. And while allies can learn from its war, Ukraine will remain the landlord of the data.
