Artificial intelligence technologies are rapidly evolving and are increasingly used both in the information war and on the battlefield.
According to Andrii Kovalenko, head of Ukraine’s Center for Countering Disinformation (CCD) and officer of the Ukrainian Defense Forces, the past month has seen several key trends in AI use:
AI-Poisoned Algorithms: Russian disinformation operations have adopted a new strategy of “poisoning” large language models (LLMs) such as ChatGPT, Gemini, and Claude. By mass-generating fake news sites, pseudo-analytics, and misleading content optimized for search engines, they feed these false narratives into the digital ecosystem, increasing the chances of AI models unknowingly repeating Kremlin propaganda—about a “Nazi Ukraine,” “American bioweapons,” or the “occupied Donbas.” This marks a new level of disinformation—not via trolls, but directly through the models themselves.
“AI is also being weaponized in cognitive warfare,” Kovalenko said. “Millions of social media posts are analyzed to map emotional states within societies. Algorithms detect anxiety, fatigue, or frustration and target those weak spots with fake narratives. This is no longer just propaganda—it’s precision psychological pressure, amplified by neural networks.”
Since early 2025, the CCD has recorded 191 Russian information operations involving AI tools.
Battlefield Autonomy: On the military front, a breakthrough in autonomous drones is imminent. The U.S. is actively testing AI-enabled drones capable of identifying targets, deciding to attack, and completing missions without human input.
“Platforms like Fury, Ghost, and Roadrunner have shown effectiveness in complex scenarios—such as GPS denial or communication jamming,” said Kovalenko. “These systems align with the broader JADC2 concept—a unified digital command network integrating air, land, sea, and cyber domains, enabling decisions to be made in seconds, not minutes.”
Ukraine is also building its own AI front:
“New-generation FPV drones equipped with computer vision can autonomously identify tanks, armored vehicles, and bunkers, avoid obstacles, and strike targets even without operator control.”
The CCD uses machine learning to detect coordinated disinformation campaigns. These models identify message anomalies, the synchronized use of bot networks, the artificial boosting of Telegram channels, and social media responses. Russian disinformation infrastructures exposed by this process are gradually being dismantled.
