EU opens investigation into X's Grok for sexual deepfake image generation
Between 29 December 2025 and 8 January 2026, Grok's image editing feature on X was misused to create approximately 3 million sexual images, including at least 23,300 of minors, according to research by the Digital Hate Response Center. The European Commission opened a formal investigation on 26 January 2026 under the Digital Services Act, alleging X failed to prevent non-consensual sexual deepfakes of women and children.
What happened
Grok, a generative AI system operated by xAI and integrated into X, was equipped with image editing functionality in December 2025. Between 29 December 2025 and 8 January 2026, the feature was misused to edit images of real people into sexual content without their consent.
According to research by the Digital Hate Response Center, approximately 3 million sexual images were generated during this 11-day period, with approximately 23,300 depicting minors. The feature was restricted to paid members on 9 January 2026 and further limited with technical restrictions on 14 January 2026.
Celebrities including Taylor Swift and former U.S. Vice President Kamala Harris were among those whose images were modified without consent. Despite the restrictions, sexual images generated through the feature remained publicly visible on X.
Reported impact
The European Commission stated that non-consensual sexual deepfakes of women and children 'demean dignity' and constitute 'violent and unacceptable' content. Indonesia temporarily blocked access to Grok, and the United Kingdom announced an investigation under its Online Safety Bill.
The Japanese government requested improvements to X to prevent inappropriate image outputs. Multiple governments and regulatory bodies identified the feature as a significant concern for online safety.
Index analysis
The incident demonstrates a failure to implement adequate safeguards before deploying image editing capabilities on a social media platform with global reach. The rapid generation of 3 million images at an average rate of approximately 190 per minute indicates ineffective content moderation and safety controls.
Governance controls
The European Commission opened a formal investigation under the Digital Services Act on 26 January 2026 to determine whether X's operator conducted adequate risk assessments and took appropriate measures to prevent illegal content circulation within the EU, particularly content depicting child sexual abuse.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.