This story contains descriptions of explicit sexual content and sexual violence.
Elon Musk’s Grok chatbot has drawn outrage and calls for investigation after being used to flood X with “undressed” images of women and sexualized images of what appear to be minors. However, that’s not the only way people have been using the AI to generate sexualized images. Grok’s website and app, which are are separate from X, include sophisticated video generation that is not available on X and is being used to produce extremely graphic, sometimes violent, sexual imagery of adults that is vastly more explicit than images created by Grok on X. It may also have been used to create sexualized videos of apparent minors.
Unlike on X, where Grok’s output is public by default, images and videos created on the Grok app or website using its Imagine model are not shared openly. If a user has shared an Imagine URL, though, it may be visible to anyone. A cache of around 1,200 Imagine links, plus a WIRED review of those either indexed by Google or shared on a deepfake porn forum, shows disturbing sexual videos that are vastly more explicit than images created by Grok on X.
One photorealistic Grok video, hosted on Grok.com, shows a fully naked AI-generated man and woman, covered in blood across the body and face, having sex, while two other naked women dance in the background. The video is framed by a series of images of anime-style characters. Another photorealistic video includes an AI-generated naked woman with a knife inserted into her genitalia, with blood appearing on her legs and the bed.
Other short videos include imagery of real-life female celebrities engaged in sexual activities and a series of videos also appear to show television news presenters lifting up their tops to expose their breasts. One Grok-produced video depicts a recording of CCTV footage being played on TV, where a security guard fondles a topless woman in the middle of a shopping mall.
Multiple videos—likely created to try and avoid Grok’s content safety systems, which may restrict graphic content—impersonate Netflix “movie” posters: two videos show a naked AI depiction of Diana, Princess of Wales having sex with two men on a bed with an overlay depicting the logos of Netflix and its series The Crown.
Around 800 of the archived Imagine URLs contain either video or images created by Grok, says Paul Bouchaud, the lead researcher at Paris-based non-profit AI Forensics, who reviewed the content. The URLs have all been archived since August last year and only represent a tiny snapshot of how people have used Grok, which has likely created millions of images overall.
“They are overwhelmingly sexual content,” Bouchaud says of the cache of 800 archived Grok videos and images. “Most of the time it’s manga and hentai explicit content and [other] photorealistic ones. We have full nudity, full pornographic videos with audio, which is quite novel.”
Bouchaud estimates that of the 800 posts, a little less than 10 percent of the content appears to be related to child sexual abuse material (CSAM). “Most of the time it’s hentai, but there are also instances of photorealistic people, very young, doing sexual activities,” Bouchaud says. “We still do observe some videos of very young appearing women undressing and engaging in activities with men,” they say. “It’s disturbing to another level.”
The researcher says they reported around 70 Grok URLs, which may contain sexualized content of minors, to regulators in Europe. In many countries, AI-generated CSAM, including drawings or animations, can be considered illegal. French officials did not immediately respond to WIRED’s request for comment; however, the Paris prosecutor’s office recently said two lawmakers had filed complaints with its office, which is currently investigating the social media company, about the “stripped” images.
