OpenAI admits AI agents accidentally posted ChatGPT users’ images online
A new disclosure from OpenAI has raised fresh concerns about the actions of increasingly autonomous artificial intelligence systems, after some of the company’s AI agents accidentally uploaded images from ChatGPT users to third-party websites.
The company said 53 images were affected, with most already removed and efforts continuing to take down the remaining files.
OpenAI said the images were transmitted by AI agents used in research after the systems sent training and evaluation data to external services without authorisation.
The affected images came from users who had agreed to have their data used to improve OpenAI’s models. The company said the data had undergone a privacy filter and could no longer be linked to the original users.
OpenAI said the incidents happened before it strengthened security measures in its research environment in August, following other cases involving AI agents operating beyond their intended limits.
The company also confirmed that some of its AI tools accessed US federal government websites, but said they retrieved only publicly available information.
OpenAI is now reviewing the past activities of its AI agents, a process it says could take several months.
Chief Executive Officer, Sam Altman, acknowledged that the company had not moved as quickly as it would have liked in reviewing and disclosing the incidents, while saying the volume of data involved made the process complex.
