Jugantor publishes AI video of child playing tug-of-war with gorilla
A video depicting a child playing tug-of-war with a gorilla has been published on the Facebook page of Dainik Jugantor. However, a verification by Dismislab reveals that the video was created using artificial intelligence (AI); it is not a video of a real incident.
Dainik Jugantor published the video from its Facebook page on August 12. The description of the post read, “Child playing tug-of-war with gorilla, and then.” In the 14-second video, a young girl is seen pulling a rope from one side of a glass panel while a gorilla, along with its offspring, are seen pulling it from the other side of the glass.
The video was also posted from multiple (1, 2, 3, 4, 5) Facebook pages with the same caption.
To verify the source of the video, conducting a reverse image search using keyframes led to a verified page of a content creator named “Yueera” on Facebook, who posted the video on August 8. The video posted on Jugantor shows an exact match with this video. Upon observation, it is seen that videos featuring gorillas and children are regularly (1, 2, 3, 4) posted from this page. Additionally, an AI label was seen attached to that specific video post.
Examining the video using the AI content detection tool Hive Moderation showed that the probability of the video being created with AI is 99.8 percent. For further verification, Dismislab tested it using Google’s AI detection tool SynthID. Google SynthID stated in its results, “Based on the analysis of the video, it was found that its visuals were created or edited using Google AI. However, no AI watermark was detected in the audio portion of the video.”
The Paris Charter on AI and Journalism published by Reporters Without Borders (RSF) gives guidelines for journalists using artificial intelligence (AI). Under this policy, several fundamental ethical standards have been defined for journalists regarding the use of AI content. One standard states that news organizations must ensure the authenticity and source of published content. Content that does not meet these criteria of genuine origin or authenticity must be considered potentially “misleading” and thoroughly verified prior to publication.
