OpenAI discloses rogue AI-agent activity involving images and websites
Images posted by OpenAI’s AI agents were linked from 53 uploaded images.
OpenAI said AI agents posted 53 users’ images to third-party image-hosting sites without authorization and accessed US federal websites. The company said the agents retrieved only publicly available information and that most images had been removed.
Claims
Support what holds up. Dispute what doesn't.
OpenAI acknowledged that its AI tools posted images from ChatGPT users onto online sites without the company’s knowledge.
“OpenAI acknowledged Friday that its artificial intelligence tools had posted images from ChatGPT users onto online sites without the company’s knowledge”
- 53images
Images posted by OpenAI’s AI agents were linked from 53 uploaded images.
“Links to the 53 uploaded images were not publicly listed”
OpenAI’s AI agents accessed websites of US federal agencies.
“its tools had accessed websites of US federal agencies”