Varanasi Police Called a Viral Video of Cops Watching a Temple Dance 'AI-Generated.' It Wasn't.

Alt News tracked down multiple independent recordings of the same performance from different attendees — including Instagram videos posted by one of the dancers herself, plus footage from other eyewitnesses (@aman_bhu_2427 and Sushila Seth) — all showing the identical performers, in matching attire, on the same stage, with the same temple backdrop ^[1].

Critically, uniformed police personnel are visible on and below the stage in every one of these independently sourced clips, not just the one Rai posted ^[1]. That rules out the police's specific rebuttal — that no officers appeared in "real" crowd footage — and rules out AI fabrication, since AI-generated video could not independently replicate matching performers, outfits, and architecture across multiple unrelated people's phone cameras ^[1][2].

The claim didn't stay confined to a press statement. Uttar Pradesh police said a case would be registered over circulation of what they called the "misleading video," and a report from The Statesman indicates an FIR was filed in connection with the footage ^[3]. The dispute quickly became politicized: the opposition Samajwadi Party's media wing seized on the original video to attack the state's BJP government, saying "obscene songs, obscene dances and vulgar performances are being held even in temple premises and religious events" ^[2], while temple authorities defended the dance as a centuries-old ritual tradition tied to the deity's symbolic wedding ^[2][3].

The video is real, not AI-generated — the police's own denial was the false claim. Alt News's cross-referencing of independently filmed footage from multiple attendees, all matching in performers, costumes, staging, and visible police presence, leaves no credible basis for the AI-fabrication claim ^[1].

This is a case of the so-called "liar's dividend": as AI-generated video becomes common enough to be plausible, authorities facing embarrassing but authentic footage can simply dismiss it as a deepfake, shifting scrutiny from their own conduct onto the people who recorded and shared it ^[1][2]. Here, an FIR over the video's circulation followed directly from a debunked claim ^[3].