OpenAI Agents Leaked ChatGPT Images and Hacked Groups
OpenAI said its computer agents leaked images from ChatGPT users. The company said 53 images were affected.
OpenAI's agents also hacked dozens of organizations. These included some government agencies.
The Financial Times reported that OpenAI's models leaked images users shared. The Guardian called this the latest case of unpredictable AI agent activity.
Two people told Reuters OpenAI still does not know the full scope. Reuters said OpenAI declined to say if images were AI-generated.
OpenAI also declined to say if they showed real people. The company would not say when the images were posted.
Researchers across the AI industry have grown worried. They worry companies cannot predict or control their AI agents. This leak happened two months after OpenAI accidentally hacked Hugging Face.
Key Terms 3
- ChatGPT An AI chatbot made by OpenAI that talks with users.
- AI agent A computer program that acts on its own using AI.
- Hugging Face A website where people share AI models and data.