Image of damage caused by leak at the Louvre is AI-generated
- Published on December 17, 2025 at 02:45
- Updated on December 17, 2025 at 02:52
- 2 min read
- By Mathilde BEAULIEU-LÉPINE, AFP France
- Translation and adaptation Masroor GILANI , AFP Pakistan
Several hundred works in the Louvre's Egyptian department were affected by a water leak in late November, but an image circulating on social media that purportedly shows the damage was generated by AI. It bears the watermark of AI chatbot Grok and contains visual inconsistencies indicative of AI-generated content.
"New crisis at Louvre: Water leak at Paris museum damages hundreds of ancient books," reads the headline of an article published by Pakistan's Geo News on December 8, 2025.
The article includes an image showing the floor of a room where artefacts are displayed covered in water. Several scattered books and what appears to be a sheet of cardboard are also lying on the floor.
"Hundreds of rare books in the Egyptian antiquities department have been damaged by a water leak, exposing chronic neglect of the historic institution's infrastructure," reads the Geo News report.
The same image was also shared in similar Facebook posts from Egypt, France and Canada, after a water leak was discovered at the Louvre on November 26 (archived link).
The leak affected "between 300 and 400 works," Francis Steinbock, deputy general administrator of the museum, told AFP on December 5. He described the damaged items as Egyptology journals and scientific documentation used by researchers that date from the late 19th and early 20th centuries.
While "extremely useful", they are "by no means unique", Steinbock said, adding there were no "irreparable and definitive losses in these collections".
The Louvre's press service told AFP on December 8 that the leak -- attributed to an accidentally opened valve in the heating and ventilation system -- occurred in a library in the Mollien wing, a space intended for researchers and closed to the public (archived link).
The service said the circulating image, however, "does not correspond in any way to reality" and does not show a real room in the museum.
AFP has previously debunked claims in French that the image shows the damage caused by the leak.
An analysis of an uncropped version of the image shared in a French Facebook post shows it bears the watermark "Grok" in the bottom-right corner, indicating it was generated using the AI chatbot.
It also contains inconsistencies indicative of AI-generated content, such as a distorted book and floor tile in the centre and right of the image. The ceiling lights also appear to be irregularly arranged.
Results from an analysis of the image on the Hive Moderation tool gave the image a 99.6 percent probability of being generated using AI.
The Louvre has also been targeted with other false claims since an embarrassing daylight heist at the museum that saw thieves make off with crown jewels worth an estimated US$102 million (archived link).
Update formatting in headlineDecember 17, 2025 Update formatting in headline
Copyright © AFP 2017-2025. Any commercial use of this content requires a subscription. Click here to find out more.
Is there content that you would like AFP to fact-check? Get in touch.
Contact us
