OpenAI Apologises To Australia Over Security Incident
OpenAI apologised to Australia after an AI agent accessed a Services Australia portal during testing, admitting authorities should have been alerted sooner.
Email: online@newsofbahrain.com
Sydney: OpenAI has apologized to Australia for an unauthorized AI-agent incursion into a Services Australia portal, admitting that it should have notified authorities earlier.
The issue occurred in June during internal testing, when an experimental OpenAI model got non-public access to the Medicare information Reporting Service and obtained internal files, credentials, and aggregate information. OpenAI stated that there was no proof of individual medical records being accessed.
OpenAI stated that it became aware of the situation in August, but only reached relevant Australian agencies in September. The business admitted that it should have published early findings earlier and promised improved controls, assistance for affected agencies, and an Australian taskforce on AI-related cyber risks.
