"God forgive, Citra Raya bridge in Cikupa, Tangerang, collapsed and disconnected," reads an Indonesian-language Instagram post shared on March 15, 2025.
It includes a video showing several scenes of what appears to be a bridge crumbling into a heavily flooded river.
Image
Screenshot of the false post, captured March 20, 2025
Similar posts also surfaced on Instagram, YouTube and Facebook after several districts in the Tangerang regency were inundated following heavy rains. At least 3,000 residents were affected, according to a report from the Tribun newspaper (archived link).
However, a representative for the local police office told AFP on March 19 that no bridge in the area collapsed during the floods. The officer described the circulating posts as a "hoax".
A Google reverse image search using the video's keyframes found the clips were taken from multiple posts on a YouTube account named "ai headline".
It describes itself as a channel "where news content is presented directly by sophisticated artificial intelligence technology" and says "the information presented is not intended as official".
All the source clips from the account bear a label that reads: "Altered or synthetic content" (archived links here, here, here and here).
The label is a feature YouTube introduced in March 2024 requiring creators to disclose to viewers when realistic content is made with altered or synthetic media (archived link).
Image
Image
Image
An analysis of the clips also shows visual errors, such as people suddenly disappearing or effortlessly walking along the collapsed structure.
While there is no foolproof method to spot AI-generated media, identifying watermarks and visual inconsistencies can help, as errors still occur despite the meteoric progress in generative AI.