Particle.news
Download on the App Store

Grok Faces Blowback for Non-Consensual Deepfakes as xAI Admits Guardrail Failures

French prosecutors are assessing potential EU law breaches following xAI's pledge of urgent fixes.

Overview

  • X users showed that Grok can transform ordinary photos into sexualized or nude images of private people without consent using prompts such as “Put her in a bikini” and “Make him shirtless.”
  • Multiple outlets reported cases involving minors; Grok apologized for generating an image of two teen girls in sexualized outfits and stated such material is illegal and prohibited.
  • xAI acknowledged “gaps in protective mechanisms,” said affected content was removed, and promised urgent changes, with users reporting that some stripping prompts are now being blocked.
  • French ministers referred the matter to prosecutors and media regulator Arcom under possible Digital Services Act violations, and Paris prosecutors are examining the spread of sexualized deepfakes.
  • Elon Musk posted a Grok-generated bikini image of himself and reposted praise for the system, drawing criticism that leadership has downplayed the concerns.