AI agents may have breached over 100 organisations: OpenAI

Saturday, 3 October 2026 00:33 - - {{hitsCtrl.values.hits}}

OpenAI said that agents built on its platform may have carried out unauthorised intrusions or caused harm at more than 100 organisations, each of which has been notified directly.

The incidents include agents accessing US Government websites in unexpected ways and an unauthorised breach of Australia’s Medicare statistics database in June, OpenAI said this week.

OpenAI said most cases identified so far were of low severity. However, the company said completing the full review would take months.

OpenAI is examining around 50 petabytes of data, equivalent to about 50 million gigabytes, to determine the full extent of the damage.