The battlefields of Ukraine are strewn with the wreckage of drones, a testament to their now-central role in modern warfare. Yet, amid this debris lies a new and lucrative resource for the defense industry. The data generated by these unmanned systems is poised to outlast the conflicts themselves, increasingly woven into the artificial intelligence (AI) frameworks that influence civilian life. Each drone flight captures thousands of data points, from imagery and video to operator inputs, creating a detailed record of how machines and humans respond to fluid and unpredictable situations.
Ukraine has begun to transform this wartime experience into a tangible asset. In January, the Ministry of Defense announced it would open up millions of data points from tens of thousands of drone flights to both military contractors and commercial enterprises. Since then, over 100 companies and the UK government have been granted access. For a nation at war, this is an expedient way to attract investment and forge partnerships. However, this move effectively turns the front line into a live training ground for AI models, exploiting the unique conditions of war that are difficult to replicate in any laboratory. Other nations and conflict zones are expected to follow suit, but the governance of this nascent industry cannot rest solely on a country fighting for its existence. The legal void must be addressed collectively by the nations and corporations involved in this sector’s growth.
Explosive Growth of Battlefield Data Use
Ukraine is not the first to use battlefield records for model development. American drones over Syria and Yemen in the late 2010s collected data that informed the first generation of semi-autonomous military hardware. The key difference now is that access to such data is being leveraged to build a broader ecosystem. The financial stakes for defense firms are enormous, as battlefield data provides vast amounts of machine experience under conditions no lab can replicate. The most valuable data for AI training comes from anomalies—moments when visibility fails, signals jam, or operators improvise. AI companies spend years and significant resources trying to capture such edge cases to make their models more robust. War produces these scenarios at a frequency that controlled testing cannot match.
This dynamic environment is what makes drone data valuable beyond the battlefield. A commercial delivery or remote-sensing drone may never face artillery fire, but it must still operate with incomplete information in a world of unpredictable human behavior. War compresses these challenges into a much shorter timeline. When processed and paired with operator records, this data turns operational logs into training sets, making combat a commercial asset. This training loop has already been observed in previous conflicts, where drone footage informed subsequent military technology. The market is expanding. Enabled Intelligence, a US company specializing in preparing data for AI training, reports it has already processed over half a million hours of Ukrainian drone footage for use in the next generation of models, with potential applications in both military and commercial systems.
Closing the Data Loop
Many of the drones that define modern warfare began as civilian technology. They have been enhanced by new, commercially available AI systems that allow inexpensive machines to operate autonomously, either individually or in swarms, adapting to changing environments. Each flight creates a record of what the system encountered, and this data is critical. Controlled lab environments used to train autonomous systems can approximate failures but cannot match the live conditions of a battlefield with real risks. For nearly a decade, military intelligence programs like Project Maven have held data from sensor-heavy systems like Predator and Reaper drones, but access was confined to the defense sector. This data was only available through restricted, classified channels for the sole purpose of developing new weapons for the same military that produced it.
Now, this experience is being shared with a much wider development network, closing the loop. Commercial technologies adapted for the battlefield are generating data that flows back into the industries from which they originated, becoming part of the data infrastructure used by both governments and the private sector. For example, drones trained in the signal-jammed airspace over Ukraine are now being used in agriculture to help farmers map and survey fields in areas lacking the cell signal required by older technologies. Other countries are likely to follow Ukraine in selling their battlefield data, and the world is unprepared for the marketplace this will create. While bad actors could acquire the data, purchase controls mitigate some risk. Intelligence operatives scrutinize potential customers’ infrastructure to prevent data from reaching enemies or malicious actors. However, training data presents a new tracing problem. Unlike commercial datasets, where planted contact details can reveal the chain of custody, the provenance of AI training data becomes obscured within the technology itself.
Another concern is the emergence of an extractive economy, where wealthier nations far from danger profit from the mortal threats faced by frontline states. This could create a market incentive for conflict to persist as an endless source of digital gold.
A Fraught New Frontier
Existing laws regulate how militaries conduct war, but they say little about what happens when combat records are stripped of their operational context, packaged as data, and licensed to companies whose products circulate far beyond the battlefield. The responsibilities of the companies that design these systems remain unresolved. Ukraine is building access controls, as mentioned in the newly signed UK-Ukraine AI agreement, but no governments are actively working on regulating what happens once data is absorbed into a model and re-enters civilian markets.
These records contain human lives. The soldiers and civilians visible in them did not consent to become training material for products that might be sold years later. Sensor data, camera footage, and coordinates from civilians fleeing a drone strike now constitute the kind of data that informs how future machines make decisions. This is a problem of consent. Individuals featured in the data—whether targets, controllers, or bystanders—become part of the training material. The autonomous capabilities built on this data do not stop at the battlefield’s edge; they move into other military or commercial systems like delivery vehicles or agricultural machinery. Errors and assumptions embedded in the data travel with the model even when it enters civilian life.
Battlefield data should not be treated as ordinary commercial material, yet no agency or regulator currently has jurisdiction over this issue. In the meantime, governments that provide access to defense data should treat it like a controlled weapons transfer, recording its origin, licensing users, and restricting onward sharing. Ukraine has begun to address this with its Avengers Labs program, which allows companies to train models on battlefield data without giving them direct access to sensitive databases. However, this only mitigates part of the problem. Governments should require disclosure when models trained on wartime material are later incorporated into civilian products. The goal of such regulation should be to make the path from combat to commerce visible.
What these companies are really mining is experience. Soldiers cannot consent to having their experience used in this way—as training data that produces a model advantage and ultimately supports a product used far from where the war was fought. The question is no longer only what technology companies can sell for use in war, but what they can extract from it. To protect against the excesses of this new industry, a regulatory system is needed that follows battlefield data wherever it goes, from combat to model to commercial product.
