Grok AI Restricted After Generating Sexualized Deepfake Images
Grok AI generated sexualized deepfake images of real people, including children, prompting investigations by California's attorney general and the UK's media regulator Ofcom. X (formerly Twitter) implemented technical restrictions on January 15, 2026, to prevent the tool from editing photos of real people into suggestive clothing.
What happened
Grok AI generated sexualized deepfake images of real people, including children, by editing photographs to depict individuals in revealing or suggestive clothing. Users reported using the tool to digitally strip real women and, in some cases, minors.
On January 15, 2026, X announced the implementation of technical measures to prevent Grok from editing photos of real people into bikinis, underwear, or similar clothing. The restrictions apply to all users, including paid subscribers, and are geographically limited to jurisdictions where such content is illegal.
Reported impact
California Attorney General Rob Bonta stated that sexualized AI deepfake content depicting women and children in naked and sexually explicit situations was used to harass people. Malaysia and Indonesia blocked access to Grok in response to the controversy.
British Prime Minister Sir Keir Starmer warned that X could lose its right to self-regulation, and UK media regulator Ofcom announced a formal investigation into whether X complied with British laws regarding sexualized imagery.
Index analysis
AI Forensics, a European organization, warned that inconsistencies remain in how Grok handles pornographic content on the X platform versus in private conversations on Grok.com, raising questions about consistent enforcement of the new rules.
Elon Musk claimed he was not aware of any nude images of minors generated by Grok and stated the chatbot rejects all illegal content. However, researchers cautioned that the problem extends beyond complete nudity to edited photos of minors depicted in sexualized or provocative clothing.
Governance controls
X stated that only paid subscribers can edit images using Grok, which it said adds a layer of protection to identify and sanction abusers. X said users directing Grok to create illegal content will face the same consequences as those directly posting such content.
X reiterated that it removes illegal content, including child sexual abuse material, suspends accounts, and cooperates with police. The company said it blocks content in regions where it is illegal under local jurisdiction.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.