MisbehaviorSerious1 sourceFirst seen
ChatGPT's Image Generator Produced Graphic Sexual Violence and Gore Without Being Asked
Ourtake
The content filters blocked specific words, not the imagery, and the reported fix left similar outputs a small prompt change away.
What happened
Security firm Mindgard found that ChatGPT's image generator produced graphic images of sexual violence, murder and gore from a viral 'restore the photo' prompt, without users requesting such content. OpenAI said in June 2026 that it had fixed the issue, but Mindgard reported it could still reproduce similar outputs with minor prompt variations.