OpenAI alerts over 100 organisations about rogue AI agent activity, check all details
Follow UsOpenAI has notified more than 100 organisations as part of an ongoing investigation into potentially unauthorised activity by its AI agents. T
OpenAI said that receiving a notification does not necessarily mean private information was accessed or that a third-party system was compromised.
The review began after the Hugging Face incident, where an AI agent was linked to a security breach.
OpenAI has notified more than 100 organisations as part of an ongoing investigation into potentially unauthorised activity by its AI agents. The review began after the Hugging Face incident, where an AI agent was linked to a security breach. OpenAI is checking historical records to identify other cases where its models may have interacted with websites or systems in unintended ways. So far, the company said it has not found another incident similar to the Hugging Face case in terms of scale or severity. However, the investigation is still underway, and OpenAI expects to uncover more cases.
SurveyOpenAI said its models sometimes need internet access to complete tasks for users. These can include researching information across websites, downloading software packages, or working with online documents. Internet access may also be used during model training and testing.
Also read: OpenAI fires three safety researchers over alleged sharing of sensitive data
However, the company said there have been cases where models used the internet in unintended ways or did not have the right restrictions in place.
“Over the last several months, we have been applying new technical and operational measures to avoid similar problems, or catch them very early,” OpenAI said. The company added that it will continue to improve these safeguards.
As part of the review, OpenAI is examining around 50 petabytes of data from its training and testing activities. The company said it is going through the records month by month to look for potentially harmful or unauthorised activity.
Also read: Temple pre-orders may start next week, Deepinder Goyal confirms six colour options
OpenAI is using AI to help analyse the large amount of data. It currently has around 7,000 GB200 and GB300 GPUs working on the review. The effort costs the company more than $500,000 per day.
As of September 26, the company had informed more than 100 organisations about activity that met its notification criteria. OpenAI stressed that receiving a notification does not necessarily mean private information was accessed or that a third-party system was compromised.
The company said it is working on standards for reporting cases where AI agents behave in ways that negatively affect third-party services.
Also read: OpenAI and Anthropic face investigation as rouge AI agents raise security concerns, check details
