KHARKIV REGION, Ukraine — Ukrainian drone crews are deploying rugged Ukraine A.I. drones that can find and lock onto Russian armor and artillery despite heavy electronic jamming along the eastern front this month. The small strike craft uses on-board image-recognition software to complete their final dive, even after losing contact with pilots — though commanders insist humans still approve every attack, Saturday, Nov. 29, 2025.
How Ukraine’s AI drones are designed to skirt jamming
Electronic warfare systems now cover large swaths of the 1,000-kilometer front, with Russian and Ukrainian units deploying truck-mounted and handheld jammers to knock conventional first-person-view drones out of the sky or blind their video feeds, sharply cutting hit rates and forcing pilots to fly closer to the line of fire. A Reuters graphic analysis of drone warfare in that year described a cat-and-mouse fight in which each new drone tactic is quickly countered by newly developed jamming methods.
To regain that accuracy, Ukrainian start-ups and volunteer engineers have been working for the past two years to train computer vision models that can run on cheap processors soldered into their strike planes. As early as late 2024, officials were speaking of “several dozen” AI guidance tools that had been fielded, allowing drones to remember the silhouette of a tank, truck or radar and to fly themselves in for the final seconds of flight — even inside electronic “domes” protecting high-value Russian targets, according to an October report that said one Ukrainian company had sold more than 15,000 copies of its targeting software. That turn was captured in a Reuters piece about Ukraine deploying dozens of A.I. systems to better enable drones to strike their targets.
Those tools expand on previous efforts. By the middle of 2024, Reuters reported that start-ups like Swarmer were rushing to develop autonomy software for swarms of drones, with engineers operating on the assumption that “soon, there is no connection to make on the front line” between operators and aircraft due to jamming. The reporting about Ukraine’s drive to develop AI-controlled war drones described simplistic lock-on functionality that could be retrofitted to low-cost FPV craft for around $150 per drone.
Now, the technology is being used regularly. In a report published Saturday, Reuters reported on how a Ukrainian pilot in the 58th Separate Rifle Brigade had deployed an AI-run targeting system to direct a drone over 20 kilometers onto what seemed to be a Russian tank, a strike he said would have been “absolutely” impossible given the type of jamming now occurring without software that could continue following the vehicle after the link with his controller was severed. That battlefield account was included as part of a feature on how Ukrainian drone pilots are turning to AI for a battlefield edge.
The AI drones targeting Ukraine are all keep-humans-in-the-loop affairs.
Ukrainian officials say Ukraine’s AI drones, although becoming increasingly autonomous, are not making life-or-death decisions on their own. Kyiv has insisted that a human touch is required for each strike, even if a drone at the end of its journey is on automated pilot, flying towards a target it had locked on to using AI. A recent report from the Center for Strategic and International Studies noted that Ukrainian drones use artificial intelligence to accelerate takeoffs, landings, and target identification, but still fly “with human oversight,” occasionally shortening the time from detection to destruction to just over 30 seconds. That study of technological evolution on the battlefield also featured Ukraine’s creation of drones guided by AI, specifically intended to counter Russian jamming.
On land, semi-autonomous swarms are a different story. Straight Arrow News recently explained how the Ukrainian company Swarmer threads drones together so that a reconnaissance craft first roams and maps a route, followed by bomber drones that choose among themselves which will strike and when, after humans designate the target. In such missions, one operator can control multiple aircraft that communicate directly with one another, making them more difficult to jam as well as relieving pressure on overworked crews, according to the report about Ukraine’s use of AI-powered drone swarms against Russian forces.
Gamification, and Ukraine AI drones for the win
And the rapid proliferation of Ukraine’s AI drones is also being fueled by incentives and battlefield necessity. Time magazine has reported that the country’s “Army of Drones” bonus system gives drone units points for verified strikes, which can be cashed in on the Brave1 online marketplace for more sophisticated hardware — such as drones that can transition into AI-assisted modes mid-flight. That gamified program, called by some “Fedorov’s killer app,” effectively turns confirmed kills and destroyed vehicles into a feedback loop that steers procurement toward systems proven to work, which increasingly means an AI-assisted targeting system. Details of the scheme appeared in a feature on how Ukraine has gamified drone warfare.
The shift has been years in the making, military analysts said. In the early days of full-scale invasion, Ukraine capitalized on cheap FPV drones and quadcopters to gain an asymmetric advantage over Russia, but hit rates plummeted as Russian jamming densified and both sides began speaking publicly about a looming “robot war.” One Reuters interactive, about how drone warfare in Ukraine is transforming the way wars are fought, relayed news of the first trials of AI-powered strikes and predicted that no one would be piloting those so-called FPV drones down the line as electronic warfare evolves.
For the time being, Ukraine’s AI drones are still largely specialized tools: They lock onto a shape, not a broader tactical view, and can still be foiled by camouflage, decoys, or bad weather. But along with swarm software and a new armed force of unmanned systems, they are changing daily combat and deepening the world’s debate about just how much control humans should give up as machines learn to hunt targets amid cacophonies of noise.
