MAIN MENU
AI-doctored protest videos reused authentic official footage; Sumit Jha resignation clip was false
The viral Sumit Jha clip reused genuine Delhi Police footage but replaced its audio to make him appear to resign and accuse the government of framing student protesters. A separate doctored Piyush Goyal video circulated during the same protest cycle, establishing recurrence of this manipulation method but not common authorship or coordination. Source-video matches, official clarifications and later police cases make the Jha claim false; AI-detector scores are supporting rather than decisive evidence.
A viral Instagram reel made Delhi Police DCP Sumit Jha appear to resign from the Indian Police Service, accuse the government of pressuring him to falsely implicate student protesters and declare that he would stand with the students instead. The resignation statement was fabricated. The visuals came from a genuine Delhi Police video posted on 23 July 2026, but the speech in the viral version was replaced with digitally manipulated, AI-generated audio.
What the initiating reel communicated
The initiating reel was posted by Avadhesh Nigam on 28 July 2026. Its caption said the officer had chosen resignation over lying, described allegations that he was being pressured to blame protesting students, and added that the claim had not yet been independently confirmed. The audio nevertheless presented a first-person resignation statement as if spoken by Jha: that he was resigning immediately, had been asked to portray students as criminals, and could not remain part of such a system.
That distinction matters. The caption retained a formal caveat, but the audiovisual artifact itself supplied a fabricated eyewitness-style statement in the officer’s own face and apparent voice. The underlying proposition was therefore not merely that rumours existed; it was that Jha himself had resigned and made the accusations heard in the clip.
The source footage says something else
Delhi Police had posted the authentic footage on 23 July. In that video, Jha discusses misinformation around the student protests, says police had identified hundreds of Pakistan-based social-media handles involved in coordinated misinformation and urges students and the public to verify material before sharing it. He does not resign and does not say that the government ordered him to falsely implicate students.
The matching setting, officer and visual sequence make the source-video comparison more probative than a standalone AI-detector score. Delhi Police publicly labelled the resignation clip fake and AI-generated on 26 July, and PIB issued a separate fact-check on 28 July. Independent checks by Alt News, CyberPeace, The Quint and others also traced the viral visuals to the police source footage and found the audio altered.
This was not the only doctored official video around the protest cycle
The wider public-interest issue is synthetic audio being attached to authentic public-official footage during a politically charged protest. AFP documented another manipulated video from the same protest cycle in which genuine footage of Union Minister Piyush Goyal was paired with fabricated audio making him appear to call for protesters to be hanged. Goyal said his remarks had been maliciously doctored using AI and reported filing a police complaint; news reports said an FIR was registered.
These two examples establish recurrence of the manipulation method around the student protests: real footage of identifiable officials was repurposed to manufacture statements that would radically change how viewers understood the officials’ positions. They do not establish that the Jha and Goyal videos had the same creator, were centrally coordinated, came from the same political actor or produced a measurable political effect. Those stronger propositions would require separate evidence.
Why authentic visuals make the manipulation consequential
The Jha clip did more than invent a resignation. It reversed the informational role of the source video. A police message warning about misinformation was converted into an apparent whistleblower-style denunciation of the government and police response. Because the face, setting and much of the video were genuine, the manipulated audio could borrow credibility from authentic visual evidence.
This is why technical detection alone is not the strongest verification method here. Automated tools cited by different fact-checkers produced varying probability scores. The more reproducible evidence is provenance: locate the earlier source video, compare the visual sequence and spoken content, check official employment or resignation records where available, and then use synthetic-media detection as supporting evidence rather than as the sole basis for a verdict.
Institutional response and limits
Hindustan Times reported on 2 August that Maharashtra Cyber Police registered a case against unknown persons over circulation of the Jha deepfake. That development shows that authorities treated the circulation as a law-enforcement matter, but an FIR is an allegation and investigative step, not a finding of who created the clip or proof of criminal liability. No evidence reviewed for this update identifies the original creator of the Jha manipulation.
The protest itself was real and politically significant, and disputes over policing, examination integrity and government accountability must be assessed on their own evidence. The falsity of the Jha resignation video does not resolve those underlying disputes. Equally, genuine controversy around the protests does not make a fabricated statement authentic.
Conclusion
The Sumit Jha resignation claim is false. Authentic Delhi Police footage was reused with fabricated audio to make Jha appear to resign and accuse the government of pressuring him to frame students. A separately documented Piyush Goyal deepfake from the same protest cycle shows that this was not an isolated use of the technique, but the available evidence supports only a pattern of similar manipulation—not common authorship, coordination or motive.
