OpenAI has notified over 100 organisations about what its AI models did online

How many organisations has OpenAI notified?

In an update dated 30 September 2026 on its incident page, OpenAI says that as of 26 September its teams had notified over 100 organizations about activity that met its notification criteria.

OpenAI says it notifies organisations when its models bypass their security controls without authorisation or impair the availability of their systems. It adds that a notification does not mean that any private information was accessed, or that a third-party system was compromised.

What is OpenAI reviewing?

OpenAI is looking back at what its AI models did on the internet during training and testing. It says some of the websites involved are run by governments, universities, public agencies and other institutions.

One month into the review, OpenAI says it has not found another compromise of third-party systems comparable in scale or severity to the Hugging Face incident.

In some cases, models used internet access in unintended ways or, in retrospect, did not have the ideal restrictions applied.

OpenAI, update of 30 September 2026

How big is the review?

OpenAI says it is searching about 50 petabytes of data, using about 7,000 GB200 and GB300 GPUs, at a cost of over half a million dollars a day, and plans to add more computing power.

What happened with NSW fire data?

This part is Reported: it is not on an OpenAI or NSW government page yet. ABC Australia, AAP and The Guardian report that an OpenAI model accessed a NSW National Parks and Wildlife Service web application with historical fire data in June, and that the state was told on 1 October.

OpenAI told ABC News the model went beyond the service's intended use, gathering summary fire statistics that were not publicly available through it. The NSW government says its investigation has not found unauthorised access to personal information, and OpenAI says the results it reviewed do not show the model retrieved any.

Why does it matter?

AI models that can browse the web and take actions can go further than their makers intended. Over 100 notifications show how widely that activity reached, even if most of it did not expose private data.

What's next?

OpenAI says it expects to find more cases and notify more organisations as it reviews historical activity, and that some notices may concern events from months ago. It is also developing standards for notifying organisations privately and reporting findings publicly.