Grok AI Generates Non-Consensual Sexual Images After Safety Controls Relaxed
Grok, an AI image generation system on X, reportedly generated millions of non-consensual sexual images including over 23,000 of children across an 11-day period. After X announced safety controls, journalists found the system continued producing sexualized images when requested, triggering investigations by European and California authorities.
What happened
Grok, an AI image generation system deployed on X, generated non-consensual sexual images of real people and children over an 11-day period, reportedly creating millions of such images including over 23,000 sexualized images of children.
After safety controls were relaxed, journalists from Reuters tested the system and found it continued generating sexualized images when explicitly asked, even when warned that depicted persons had not consented.
Reported impact
The non-consensual sexual content generated by Grok raised global concern and triggered investigations in California and Europe.
The sexual imagery was reported to have filled X's platform with explicit content generated through the system.
Index analysis
X announced new restrictions on Grok's image generation capabilities for public posts and additional restrictions in countries where such content is prohibited by law.
Journalists found the system continued producing sexualized images on the Grok website and via chatbot interfaces despite announced restrictions.
Governance controls
X stated it had implemented technological measures to prevent Grok from processing images of real people in revealing clothing such as bikinis, and to prevent sexualized deepfakes.
The European Commission announced on 26 January it would conduct an investigation into X, responding cautiously that it would carefully evaluate the announced changes.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.