Airbnb Host Used an AI-Generated Photo to Demand $1,700 — the Guest Caught It With a Watermark Detector

According to a report by the foreign outlet PC Gamer, a user named FurinaDeFontaine recently posted on Reddit saying that after checking out of a guesthouse booked through Airbnb, the host used an AI-generated photo of a supposedly “leaking toilet” to try to extort US$1,700 (about 11,414 yuan at current exchange rates) in “repair costs.”

According to FurinaDeFontaine, the host in question is said to belong to a large Australian guesthouse operating company. Right after the guest checked out, the host sent the supposed photo of the “leaking toilet.” But FurinaDeFontaine quickly noticed obvious problems with it: “There’s actually another floor below the toilet in question. If there really were inches of standing water, the whole floor below should have shown obvious water damage — that doesn’t make sense.” He therefore concluded that the photo was very likely fabricated.

Afterwards, FurinaDeFontaine used Google Gemini to analyse the photo and found that it contained Google’s SynthID digital watermark, which indicated that the host had in fact used Gemini to generate the fake photo.

In the end, FurinaDeFontaine contacted Airbnb’s official customer service, which determined that the host’s compensation claim did not qualify and penalised the host’s account.

This incident shows once again that before leaving a guesthouse or hotel, people should immediately photograph the room, the bathroom and other areas prone to disputes, and keep video records, so as to provide more complete on-site evidence if a damage claim is disputed. At the same time, people should not readily trust all kinds of “photos” online: many may already have been processed with AI and may not represent the actual situation.