The Deepfakes Analysis Unit (DAU) analysed a video that apparently shows Gen. Upendra Dwivedi, India’s Chief of Army Staff, supposedly making a statement about India having lost 250 soldiers to Pakistan and the monitoring of its military assets by Chinese and Pakistani satellites. After putting the video through A.I. detection tools and getting our expert partners to weigh in, we were able to conclude that an A.I.-generated audio clip was spliced with real audio to manipulate the video.
The two-minute-and-11-second video in Hindi was discovered by the DAU on X, formerly Twitter, during social media monitoring. The video was posted from several accounts on the platform, although with different text.
One of the accounts posted the video on July 28 with the accompanying text in English, which read: “Upendra Dwivedi Ji admits to losing more than 250 soldiers in war against Pakistan. Pakistan was aware of our every single movement due to Chinese satellite.” We do not have any evidence to suggest if the video originated from any of the accounts on X or elsewhere.
The fact-checking unit of the Press Information Bureau (PIB), which debunks misinformation related to the Indian government recently posted a fact-check from their verified handle on X, debunking the video attributed to General Dwivedi.
In the video, the general is captured in a medium close-up. He appears to be standing at a podium in uniform, seemingly addressing an audience as microphones bearing logos— resembling those of media houses— can be seen placed before him. He seems to be glancing downwards from time-to-time, as if referring to his speech notes.
Black text graphics with his name and designation set against a white background appear superimposed to the left of the video for the initial seven seconds and then disappear. A logo resembling that of The Indian Express, an English-language daily and digital publication, is visible in the top right corner of the video frame throughout. Captions visible at the bottom of the video frame display a rough English translation of the entire audio track as it plays out.
The backdrop in the video, though seemingly dark, appears to be animated as white-coloured smudges pop up a few times with one especially visible near the logo. A faint impression of the Indian national flag surfaces in the backdrop during the last 10-seconds of the video.
The video quality is not sharp, however, it is clear enough to discern that the audio seems to be synchronised with the general’s lip movements for the most part. The exception is a 14-second segment with a noticeable lag between the audio and video track, resulting in the general’s lip-sync being inconsistent. In the same segment his upper set of teeth seem to disappear and reappear between frames and the inside of his mouth looks unnaturally dark.
On comparing the voice attributed to the general with his recorded videos available online there is a striking similarity in the voice, diction, and overall delivery but for that 14-second segment, which sounds slower, scripted, and differs in pitch and intonation.
Another rather unusual sound in the video, one that resembles that of a chime, can be heard at the 25-second mark. It is that characteristic ding heard on a YouTube channel encouraging viewers to subscribe.
We undertook a reverse image search using screenshots from the video and traced the general’s clip to this video published on July 26, 2025 from the official YouTube channel of The Indian Express. The clothes, body language, and backdrop of the general in this video and the one we reviewed are identical.
When we compared the two videos we noticed that the audio tracks in both the videos are primarily in Hindi with a few words in English. The audio content in the videos is identical but for that 14-second segment that we pointed to above. The chime sound heard in the doctored video is part of the original audio track.
It seems that a segment from the source video was lifted and the fake audio track was inserted to create the doctored version. There is no visible transition or jump cut in the manipulated video around that particular 14-second segment.
The source video has the logo of The Indian Express in the top right corner of the frame and a subscription button in the bottom right corner. The text graphics with the general’s name and designation as well as the captions with the English translations seen in the doctored video have been lifted from the source video. The white smudge near the logo and the faint impression of the Indian tricolour too are part of the source video.
This video, yet again, follows the trend that we have observed in the doctored videos that have surfaced in the wake of the recent military escalations between India and Pakistan wherein top defence officials and senior ministers are targeted. Such videos are typically being created by splicing authentic audio with synthetic audio or using a synthetic version of an original audio track to script a fake and harmful narrative about the supposed losses of personnel and equipment suffered by India. The DAU has debunked a few such videos, you can read our related assessment reports here.
Shared below is a table that compares parts of the audio tracks from the original video featuring the general with the doctored one to give our readers a sense of how similar the audio tracks are and how that can mislead them. We, of course, do not want to give any oxygen to the bad actors behind this content.
To discern the extent of A.I. manipulation in the video under review, we put it through A.I. detection tools.
The voice tool of Hiya, a company that specialises in artificial intelligence solutions for voice safety, indicated that there is a 10 percent probability of the audio track in the video having been generated or modified using A.I.
Hive AI’s deepfake video detection tool highlighted a few markers of A.I.-manipulation in the video track. Most of the markers were in the 14-second segment with the suspicious audio that we pointed to above. However, their audio detection tool indicated that the entire audio track is “not A.I.-generated”.
We also ran the audio track through the A.I. speech classifier of ElevenLabs, a company specialising in voice A.I. research and deployment. The results that returned indicated that it was “very unlikely” that the audio track used in the video was generated using their platform.
When we reached out to ElevenLabs for a comment on the analysis, they told us that they conducted a thorough technical analysis but were unable to conclusively determine that the reported synthetic audio originated from their platform.
To get further expert analysis on the video, we escalated it to the Global Online Deepfake Detection System (GODDS), a detection system set up by Northwestern University’s Security & AI Lab (NSAIL). The video was analysed by two human analysts, run through 22 deepfake detection algorithms for video analysis, and seven deepfake detection algorithms for audio analysis.
Of the 22 predictive models, four gave a higher probability of the video being fake and the remaining 18 gave a lower probability of the video being fake. Six of the seven predictive models gave a high confidence score to the audio being fake and the one remaining model gave a low confidence score to the audio being fake.
In their report, they highlighted several visual discrepancies in that specific 14-second segment that we pointed to above. They echoed our observations about the inconsistent lip-sync in that segment. They noted that the general’s jaw and mouth movements appear exaggerated, and his teeth seem to disappear and reappear, also something that we observed.
They mentioned that during that segment, the general’s face overlaps with his collar at one point, but instead of the collar appearing behind the face, it seems to subtly intersect with his lower jaw. They added that there is an observable lighting change surrounding the subject in that segment compared to the rest of the video.
In conclusion, the team stated that the video is likely generated via artificial intelligence.
To get another expert to weigh in, we escalated the video to our partner GetReal Security, co-founded by Dr. Hany Farid and his team, they specialise in digital forensics and A.I. detection. Referring to the 14-second segment that we flagged as suspicious, the team stated that the audio in this segment exhibits signs of manipulation.
They noted that the vocal mannerisms of the subject change subtly during that time period and there is a change in the background noise characteristics of the audio spectrogram. They added that the acoustic characteristics also change during the same time period, making it seem that the suspicious segment was recorded in a smaller room or possibly synthetically generated.
They pointed to evidence of duplication at the end of the suspicious segment, as they heard the exact same breathing and mouth noise artefacts twice in a row in short succession. They shared that as this happens at the boundary where the suspicious segment ends, it is further evidence of the use of audio editing software.
On the basis of our findings and expert analyses, we can conclude that synthetic audio was added to the general’s real audio to peddle a false narrative about India suffering the loss of 250 soldiers to Pakistan in the recent India-Pakistan conflict.
(Written by Debraj Sarkar and Debopriya Bhattacharya, edited by Pamposh Raina.)
Kindly Note: The manipulated audio/video files that we receive on our tipline are not embedded in our assessment reports because we do not intend to contribute to their virality.
You can read below the fact-checks related to this piece published by our partners:
Fact Check: Did India Army Chief Admit 250 Soldiers Were Martyred In Recent Conflict With Pakistan?
Did Indian Army Chief Admit to Losing 250 Soldiers in War Against Pakistan? No!
