Russian Families Turn to AI to Recreate Voices of Soldiers Killed in Ukraine
Across Russia, families who lost sons, husbands, and fathers in the Ukraine war are using artificial intelligence services to recreate the voices and digital likenesses of the deceased. The practice, reported by independent media outlets throughout early 2026, involves feeding old voice messages, videos, and photographs into AI platforms that generate interactive simulations of the dead.
AI Voice Cloning Services See Rising Demand in Russia
Several Russian-language AI platforms now offer “digital resurrection” services priced between 3,000 and 15,000 rubles. Users upload audio recordings—often salvaged from Telegram voice notes sent from the front—and the system produces a synthetic voice that can speak new sentences. Some services add animated avatars synced to the cloned voice. One Moscow-based developer, speaking anonymously to the independent outlet Meduza in March 2026, said demand rose sharply after the Kremlin’s autumn 2025 mobilisation wave sent thousands more conscripts to the front.
The technology itself is not new. Voice synthesis tools from companies like ElevenLabs and Respeecher have powered entertainment and accessibility applications for years. What shifted in Russia was the emotional weight attached to each generated phrase—a mother hearing her dead son say “I love you” in a voice that no longer exists anywhere else.
How the Reconstruction Process Works
The reconstruction pipeline typically starts with as little as 30 seconds of clean audio. Machine learning models analyse pitch, rhythm, and speech patterns, then build a voiceprint. Text-to-speech engines apply that voiceprint to new words typed by the user. For video avatars, facial mapping tools extract expressions from old photographs and animate them in sync with the generated audio. The result is a short clip in which a deceased soldier appears to speak directly to the camera.
Quality varies considerably. Short clips with strong source material can sound convincing. Longer outputs often slip into an uncanny flatness—correct syllables, wrong intonation. Families who spoke to the BBC’s Russian service in April 2026 described the experience as both a comfort and a wound. One woman from Novosibirsk said she plays her son’s AI-generated voice every morning, then spends the rest of the day crying.
Mental Health Experts Flag Risks of Prolonged Use
Psychologists who study grief have cautioned that AI resurrection may interfere with natural mourning. Dr. Elena Petrova, a clinical psychologist based in Tbilisi who treats displaced Russian families, told Novaya Gazeta Europe in May 2026 that she had seen three patients who formed daily rituals around AI conversations with dead relatives. “The brain does not process loss when the voice never stops,” Petrova said. “It keeps the wound open indefinitely.”
Unlike traditional memorials—photographs, letters, grave visits—AI simulations respond. They generate new content. That interactivity, researchers warn, can create a compulsive loop. A 2025 study from the University of Oxford’s Digital Ethics Lab found that prolonged engagement with “griefbots” correlated with higher rates of prolonged grief disorder in a sample of 400 users across four countries.
As digital tools become more embedded in daily routines, questions of privacy and security overlap with emotional wellbeing in ways few regulators anticipated. The data fed into resurrection platforms—intimate voice recordings, personal photos—sits on servers with varying protections.
Russian State Media and Tech Industry Response
State-controlled Russian media have largely ignored the trend, framing technology coverage around national self-sufficiency and import substitution. Independent tech journalists note a split: smaller Russian AI startups quietly market resurrection features through Telegram channels and VKontakte groups, while larger firms avoid association with the war’s human toll. No specific regulation governs digital resurrection in Russia as of mid-2026.
Outside Russia, the ethical debate has moved faster. The European Union’s AI Act, which took effect in phases through 2025 and 2026, classifies emotion-manipulating AI systems as high-risk. Digital resurrection tools fall into a grey zone—they manipulate emotion, but at the user’s request. The Council of Europe has scheduled consultations on the topic for late 2026.
Meanwhile, the technology continues to spread. Online tutorials in Russian walk users through setting up open-source voice clones without paid services. A single YouTube video from January 2026, viewed over 800,000 times, demonstrates the process using only a laptop and free software. In a country where official casualty figures remain classified, families are finding their own ways to hear the dead speak again. Whether that helps them heal or traps them in grief is a question no algorithm can answer.