OpenAI works to stop ChatGPT generating 'sex crime scene' images

Researchers from British AI security startup Mindgard found that ChatGPT could be prompted to generate sexualized and violent images. OpenAI added safeguards after being contacted by the BBC, but the problematic prompts still produced concerning content with small alterations. Mindgard's founder Peter Garraghan described the images as 'very gruesome, sometimes sexualised'. Safety researcher Jim Nightingale said he was 'shaken, and in tears' by the images.