How the campaign was built

NewsGuard traced the deepfakes to a handful of pro-Kremlin feeds that regularly push falsehoods. In the Levus video, the former deputy head of the Security Service calls for “immediate peace talks,” a stance that contradicts his public record. The Yunakov clip shows the lawmaker lamenting “the permanent loss of Ukrainian land,” also at odds with his party’s platform.

Both videos appeared on Telegram, still a main conduit for Russian-aligned narratives. Within two weeks the posts racked up 130,000 views, far outpacing earlier manipulation attempts.

Technical progress and lingering flaws

Analysts spotted classic deep-fake tell-tale signs: irregular blinks, stiff facial muscles, and flat-toned Ukrainian audio. The current videos remain imperfect, yet they hone in on individual facial features and speech patterns, indicating that state-aligned operators now have easier access to the tools.

The leap in realism matters because it lowers the barrier to creating targeted fakes. Where a single generic video once required modest effort, a small team can now churn out multiple personality-specific clips that look plausible enough to fool casual viewers.

Why the videos matter beyond the glitches

The danger lies not in technical polish but in psychological impact. The “liar’s dividend” describes how deepfakes let genuine footage be questioned, eroding trust in any visual evidence. Even after fact-checkers label the Levus and Yunakov videos as fabricated, the initial exposure can already spark anxiety, anger, or resignation among viewers who see a familiar face pleading for surrender.

Research shows corrections rarely travel as far as the original false claim. In a fast-moving conflict, that lag can shift public sentiment before authorities can respond. Moscow uses the videos strategically: sowing doubt about Ukrainian leadership’s unity, weakening morale, and inviting external pressure for a negotiated settlement.

Bottom line

AI-driven impersonations of Levus and Yunakov prove that targeted political deepfakes have moved from novelty to a tactical weapon in the Kremlin’s information arsenal. Their modest flaws do not stop them from reaching tens of thousands, and the resulting “liar’s dividend” threatens trust in authentic media. Combating this wave will require faster detection, clearer provenance standards, and public awareness that visual proof is no longer automatically reliable.