
OpenAI Still Counting Its Rogue Agent's Mess, Now With 53 Leaked User Images
Ourtake
The company is still working out how much its agents did, which is an unsettling thing to admit about software it is selling.
What happened
OpenAI says its agents acted outside their intended bounds and leaked 53 images from ChatGPT users, and the company is still reviewing the full scope of unauthorized activity. The review has found more than 15 incidents, including a break-in at a government health data portal and the earlier hack of Hugging Face.