Nearly 1,000 AI-generated songs advancing pro-Russian narratives about the Ukraine war have flooded major streaming platforms, according to new research from a Ukrainian nonprofit.
Terrecon, an organization focused on threat intelligence, catalogued 981 tracks created with artificial intelligence that it classified as war propaganda favoring Russia’s invasion. The songs span genres from ballads to heavy metal, with some glorifying Russian soldiers and containing ethnic slurs targeting Ukrainians. “Power of Artillery,” a Russian-language rock track about artillery systems, ranks among the most popular.
Most of the datable songs appeared after 2024, a timeline that aligns with the broader rollout of AI music tools like Suno and Udio. Those generators slashed the time and money needed to produce music to near zero. Daria Verbytska, a Terrecon founder, drew a parallel to how Russia adapted drone production. “They have found new ways to create cheaper drones and use them on a massive scale,” she said. “It’s like that with music: creating massive propaganda with the support of AI.”
The tracks borrow heavily from Soviet-era war music themes but package them in contemporary styles. One song, “Soldier’s Daughter,” mourns a fallen soldier while framing his sacrifice as noble. Direct Kremlin involvement remains unproven. Still, Moscow’s Internet Development Institute, which works closely with the government on patriotic initiatives, recently claimed support for more than 350 war-related artistic projects, per Russian media.
Researchers also tested four AI music generators: Suno, Udio, Google Flow Music, and Singapore-based Mureka. They entered prompts in Russian and English requesting violence and derogatory content. Some requests got blocked, but the team still produced tracks resembling those in their database. The moderation filters, according to Terrecon’s findings, flagged banned words rather than the prompts’ broader intent.
Streaming platforms now face pressure to respond. Spotify introduced labels for AI-generated songs in September and stopped recommending them. Deezer and YouTube require certain AI content to be disclosed. Deezer reviews potential violations case by case, while SoundCloud stated that content inciting or glorifying violence has no place on its service. Yet the sheer volume of material, combined with automated systems struggling to catch offensive language across multiple languages, leaves significant gaps.
The phenomenon exposes how quickly generative AI can industrialize propaganda. Streaming services and AI companies alike must now reckon with rules built before machines could mass-produce politically charged music at this scale.
