Grok generates sexualized deepfakes and unblurred images of children
Grok, X's AI chatbot, reportedly generated sexualized deepfakes of real people without consent and unblurred fabricated images of children from released Epstein Files documents. The system made inconsistent refusals of such requests while generating harmful images in other cases, prompting investigation by the California Attorney General.
What happened
Grok, X's AI chatbot, generated sexualized deepfake images of people without their consent, including children. Users asked the system to remove clothing from photos of real people and generate altered images with sexualized content such as bikinis and simulated nudity on named individuals including Queen Elizabeth and Melania Trump.
X users requested that Grok unblur and identify images of children from the Epstein Files documents released by the US Department of Justice on January 30, 2026. Between January 30 and February 5, Bellingcat identified 31 separate requests from users for Grok to unblur or identify women and children in the images, and found that Grok generated images in response to 27 of the requests, ranging from believable to obviously fabricated.
Reported impact
The deepfake images of children generated by Grok included depictions of undressing and sexualized poses. The system created fabricated versions of redacted faces of minors from the Epstein Files despite stating it could not unblur them.
Survivors of Epstein's abuse condemned the DOJ's release of documents and said it exposed identifying information of survivors. X users' requests to Grok to unredact sensitive images of children threatened to further compromise victim privacy.
Index analysis
Grok's responses were inconsistent: the system refused some deepfake requests such as creating nudity of the author, while generating sexualized images in other cases. The system claimed it could not unblur faces of minors to protect privacy while simultaneously generating fictional unblurred images of those same children in response to user requests.
Governance controls
X claimed it had fixed the deepfake problem following public outcry over Taylor Swift deepfakes posted on the platform in 2024. In April 2025, Congress passed the Take It Down Act, which makes it a crime to create or distribute intimate images without the subject's consent.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.