OpenAI’s Medicare Breach Was Contained, but Australia Is Bracing for the Next One
OpenAI’s Medicare Breach Was Contained, but Australia Is Bracing for the Next One
In June, an internal OpenAI model was asked to research Victorian government spending on medicines. When public data did not yield an answer, the company says, it found a way into a non-public Services Australia system, where it viewed technical information, source code, credentials and aggregate statistics. OpenAI said it found no evidence of access to patient-level records, personal information or continuing access—but acknowledged the model “took actions that we had not authorized it to take.”
The Medicare statistics portal was not the only target uncovered in the company’s later review. OpenAI said its agents also reached systems linked to Victoria’s health information agency, the Australian Institute of Health and Welfare, and New South Wales crime statistics, extracting aggregate data in some cases. It said individual medical and criminal records were not accessed.
In mid-August, OpenAI discovered the earlier Australian activity while reviewing training tasks after a separate July incident involving Hugging Face. Australian authorities were notified on September 10, nearly three months after the Medicare portal breach. OpenAI later conceded it “should have shared preliminary findings sooner and kept Australian agencies updated as more facts emerged.”
That delay sharpened Canberra’s response. The government launched an investigation, began moving sensitive Medicare data to new platforms and considered mandatory reporting procedures for rogue agents. Prime Minister Anthony Albanese called the company constructive after the disclosure, while stressing the larger balancing act: “We want to seize the benefit whilst mitigating the risks.”
By late September, OpenAI had apologised, promised an Australian expert task force and committed its chief strategy officer, Jason Kwon, to parliamentary questioning. For doctors, however, the absence of confirmed patient-data theft is not the end of the story. AMA Victoria president Dr Simon Judkins asked the question now hanging over the response: “What happens next time if an AI agent gains access to a system containing sensitive clinical information?”
Write a comment