The Deepfakes Analysis Unit (DAU) analysed a video apparently featuring Sundar Pichai, Google’s chief executive, and Pooja Shali, a journalist with India Today, an Indian news channel, purportedly promoting a financial investment platform being linked to Google. India Today clarified in a post on X that the video is fake and that they did not air any such news report. After running the video through A.I.-detection tools and seeking expert analysis from our partners, we were able to conclude that the video was fabricated with synthetic audio.
A Facebook link to the one-minute-15-second video in English was escalated to the DAU tipline for assessment. Even though the video is not accessible through that link anymore, the content may still be available on other platforms.
The video opens with Ms.Shali apparently speaking to the camera in a studio setting. A female voice recorded over her video track announces the launch of “Google Invest” — a supposed financial platform that would help Indian citizens earn “passive income”. The video then switches to a split screen with Shali in one frame and Mr. Pichai in another followed by a clip that features only Pichai. The audio with his clip talks about the so-called A.I.-powered investment platform that offers high returns even on small investments.
A logo similar to India Today’s can be seen in the top right corner of the video frames that show Shali and Pichai together. In those few seconds, a super carrying the text, “A revolution from Google for all the citizens of India” is visible across the top of the video; and the words “Big Story” arranged vertically, appear between the space separating Shali’s and Pichai’s visuals. A logo resembling that of “Bandhan Mutual Fund”, an India-based mutual fund house, is noticeable on the laptop facing Shali.
The last clip in the video features another female. She can be seen holding a microphone bearing a logo resembling that of India Today; the same logo is also visible in her backdrop along with a partly visible logo similar to that of Aaj Tak, a Hindi news channel, both channels are owned by the same media group. Multiple monitors are also noticeable in the backdrop, making it seem like a newsroom setting.
The video ends with a black screen emblazoned with the words “Registration is now open” and an arrow pointing downwards.
In August, the DAU had debunked another video featuring Pichai, in which a platform with the exact same name was being peddled. A.I. voice was used with an original video clip featuring him to fabricate the content. No such investment platform has been promoted by him or the company he leads.
In the video being analysed through this report, Pichai appears to be speaking to the camera from an office-like setting unlike the other video. The message and tone of the audio with his visuals is similar in both the videos but the audio script is not identical.
The lip-sync of all the three subjects in the video is imperfect. The chin and neck area of Shali and the other lady appear blurred compared to the rest of their face. In their respective segments, in some frames their jawline and chin appear to be changing contours. Shali’s lip movements seem unnaturally fast. The teeth of the other lady seem to lack natural shape, her lower set of teeth seem sparkling white.
The voice attributed to Shali sounds somewhat like hers when compared with her recorded videos available online, but lacks the natural inflection or pauses noticeable in her delivery. The accent of that voice sounds odd and the pronunciation of “Sundar Pichai” is especially peculiar.
The audio track accompanying the frames featuring the other lady sounds hastened, robotic and has an accent that sounds foreign. That voice too pronounces “Sundar Pichai” in an odd manner. We could not find any recorded videos of hers to make a comparison between her actual voice and the one heard in the video.
Pichai’s mouth movements appear to be puppet-like with a noticeable quivering from his mouth area to his chin, throughout the segment featuring him. The shape and visibility of his teeth is inconsistent, ranging from blurred to extra bright; and his eye movements also look odd in some frames.
The audio track accompanying Pichai’s video track bears some similarity to his voice but sounds hastened, scripted, and monotonic. It lacks the accent and pauses heard in his recorded interviews; overall it does not match his characteristic delivery.
A reverse image search using screenshots from the video helped establish that the video is composed of unrelated clips that have been stitched together. We were unable to locate the source of the clip featuring the lady seen in the last segment of the video.
Pichai’s segment led us to this video published on July 30, 2020, from the official YouTube channel of The Straits Times, a Singapore-based daily. We traced two videos — this and this — posted from India Today’s verified X handle on Oct. 23, 2024, which are almost identical to the clip of Shali’s seen in the video. The difference, however, lies in details such as the positioning of the lapel microphone, placement of a glass of water and mobile phone on the table in front of her. It is possible that the exact clip used in the doctored video was from a news show anchored by Shali on Oct. 23.
The clothing, backdrop, and body language of Pichai in the video that we traced and the doctored video are an exact match. The language in both videos is English, however, the content is totally different. There is no mention of any investment platform in the original video.
The manipulated video seems to have used zoomed in frames from the original video featuring Pichai, editing out the objects visible in the background and foreground.
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 returned results indicating that there is a 99 percent probability that an A.I.-generated audio track was used in the video.
Hive AI’s deepfake video detection tool indicated that the video was manipulated using A.I. It pointed out markers of manipulation in multiple frames featuring Pichai as well as in the last segment featuring the lady. Their audio tool indicated that most of the audio track was manipulated with A.I., but for the last 10-seconds.
The deepfake detector of our partner TrueMedia indicated substantial evidence of manipulation in the video. The “A.I.-generated insights” offered by the tool provide additional contextual analysis by stating that the video transcript uses overly promotional language that is common in scams and lacks the formal tone associated with official announcements.
In a breakdown of the overall analysis, the tool gave a 91 percent confidence score to the “face manipulation detector” subcategory that detects potential A.I. manipulation of faces, as in the case of face swaps and face reenactment. It gave a 78 percent confidence score to the “deepfake face detector” subcategory that uses a visual detection model to detect faces and to check if they are deepfakes. It also gave a 67 percent confidence score to the “video facial analysis” subcategory that analyses video frames for discrepancies in facial features.
The tool gave a 100 percent confidence score to the “A.I.-generated audio detector” subcategory that detects synthetic audio. It also gave a 99 percent confidence score to the subcategories of “voice anti-spoofing analysis” and “audio authenticity detector”; both analyse the audio for evidence of it being created by an A.I. audio generator or cloning.
For a further analysis on the audio track we put it through the A.I. speech classifier of ElevenLabs, a company specialising in voice A.I. research and deployment. It returned results indicating that it was highly likely that the audio track featured in the video was generated using their software.
To get another expert view on the manipulations in the video, we reached out to our partners at RIT’s DeFake Project. Saniat Sohrawardi from the project said that scam videos like these tend to follow the same narrative and use the same technology.
Mr. Sohrawardi noted that the video seems to have used the wav2lip technology — a speech-to-lip generation code repository available online for video generation. He added that his observation was based on the fact that this technology has easier giveaways than any newer lip-sync techniques.
He also highlighted visual artefacts of manipulation in the video, including “toading” and glitches between the lips. By toading he refers to a subject’s throat area inflating and moving like that of a toad when it breathes. The odd lip movements in the video could be a result of the algorithm failing to extract proper landmarks or structural details of the mouth.
To get another expert to weigh in on the video, we escalated it to our partner Validia, a San-Francisco based deepfake cybersecurity service. They used their proprietary software to check the authenticity of the video and audio track associated with Pichai.
They retrieved a clean sample of the voice purported to be Pichai’s from the video we escalated to them. Following which they generated a heat-map to compare the retrieved audio with a real voice sample of Pichai’s.
.png)
The team at Validia stated that their heat-map analysis showed significant similarities between Pichai’s real audio and the fake audio attributed to him in the video escalated by us. However, they noticed numerous discrepancies as well in their similarity check. They assessed that text-to-speech software was used to create the A.I. audio, which is apparent from incorrect pronunciation of certain words.
The team added that when they ran the video through their facial forgery detection model, they saw high probabilities of fake. They further noted that visual artefacts are visible near Pichai’s lips throughout the video.
They concluded that the video segment featuring Pichai used lip-sync deepfake technologies and a poor audio deepfake. And the video context is very unlike Google’s current business model.
On the basis of our findings and expert analyses, we can conclude that the video featuring Shali and Pichai was manipulated using A.I.-voice. It appears to be yet another scam being peddled by falsely linking the content to a top business leader.
(Written by Debopriya Bhattacharya and Debraj Sarkar, edited by Pamposh Raina.)
Kindly Note: The manipulated video/audio 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:
Viral Videos Of Sundar Pichai and Sudha Murty Endorsing Investment Platforms Found To Be Deepfakes
