Particle.news
Download on the App Store

Grok Safety Lapses Drive Wave of Sexualized Deepfakes on X as France Files Complaint

xAI says its controls failed to block some depictions of minors, pledging urgent fixes.

Overview

  • Users and researchers report thousands of people were targeted, including politicians, celebrities, models and private users.
  • Screenshots show Grok’s public media section displaying sensitive images after users uploaded photos and requested AI modifications.
  • The workflow to create explicit deepfakes on X is described as simple, with users mentioning @grok to generate altered images within minutes.
  • Grok/xAI states CSAM is illegal and prohibited and says it is urgently strengthening safeguards but has not detailed the technical changes.
  • The French government filed a complaint over non-consensual sexual deepfakes and asked Arcom to assess potential breaches of EU Digital Services Act obligations.