A video is going viral on social media, purportedly as a news segment presented by Aaj Tak anchor Shweta Singh on a betting app called Aviator, with Virat Kohli giving his testimony on how he earned money on the app.
BOOM found this video to be fake; manual inspection of the video, along with analyses by various detection tools, have revealed that the video is a deepfake - with voice clone and lip-sync algorithms being used to manipulate the video.
The video was shared by a Facebook user without a caption.
Click here to view an archive of the above post.
Fact Check
We ran a view keyword searches to find out more about whether Kohli promoted such a betting app, and found that there are no credible reports of him ever doing so. However, we found several instances of Kohli being deepfaked to promote scam betting apps.
We did a reverse image search of some of the keyframes showing Kohli's face, which led us to a video of him promoting a health product by Herbalife.
We did a side-by-side comparison of the frames of the viral video to the one posted by Herbalife, and found them to be exactly same - it has the exact same background, with Kohli wearing the same t-shirt.
After viewing the entire video of Kohli promoting the Herbalife product, we could not find any instance of him talking about a betting app in the video. This suggested that the viral video might be a deepfake.
We ran the video through the deepfake detection tool by TrueMedia, which showed a high possibility of it being a deepfake.
TrueMedia's analysis found substantial evidence of the audio being generated using artificial intelligence, while also finding evidence of manipulation of face using AI.
We also ran the video through Deepfake-o-meter, a deepfake detection by University at Buffalo's Media Lab - which further confirmed that a lip-sync algorithm has been used to manipulate the movement of the lips in the faces seen in the video.
This provides considerable evidence suggesting that the audio has been AI-generated, and the lips have been synchronised to the audio using AI as well.
