ChatGPT can be made to generate sexualised and violent images, researchers find (opens original article in a new tab)

Researchers discovered that ChatGPT can be manipulated to generate sexualised and violent images, prompting OpenAI to add safeguards, though concerns remain about AI's ability to produce harmful content.

• Researchers found that ChatGPT can generate sexualised and violent images with altered prompts
• OpenAI claims to have added safeguards against such content but researchers say the issue persists
• Mindgard's research highlights ongoing challenges in preventing AI from generating harmful content
• AI models lack understanding of intent, context, or propriety, making content control difficult
• valence:-1
• tldr:Researchers discovered that ChatGPT can be manipulated to generate sexualised and violent images, prompting OpenAI to add safeguards, though concerns remain about AI's ability to produce harmful content.