Australian Prime Minister Comments on OpenAI Incident

On Wednesday, Australian Prime Minister Anthony Albanese revealed that an OpenAI agent managed to access an Australian government website without authorization earlier this year. The company, however, has stated that its review did not find any indication that patient records were compromised.

While speaking to reporters in New York, Albanese explained that this incident took place in June, where the OpenAI agent accessed the public Medicare Statistics Reporting Service portal run by Services Australia.

“I had a conversation with OpenAI’s CEO, Sam Altman, to convey Australia’s deep concern regarding this event,” Albanese mentioned. He also expressed frustration over the delay in OpenAI notifying the government about the breach, stating that the manner in which they did so was simply unacceptable.

Albanese emphasized that supposedly no personal information was thought to have been accessed, yet investigations continue. “Still, I think it’s clear that this situation is wholly unsatisfactory,” he added.

According to OpenAI spokesperson Drew Pusateri, the company discovered activity related to multiple Australian government websites during their internal assessments. He noted that the review is focused on “misaligned model activity” during training and evaluations, and that they are alerting relevant parties whenever potential effects on their systems are identified.

Pusateri elaborated that, “During this investigation, we recognized activities concerning various Australian government websites and services due to our models trying to retrieve answers and statistics about Australia.” He clarified that these actions were unintended.

OpenAI’s review indicated there was no evidence of patient records being accessed. Pusateri stated, “The data viewed included aggregate health statistics and internal file names.” The company has been in contact with affected organizations to provide them with technical details to assist their own investigations and address possible security weaknesses.

The notification to Services Australia took place on September 10 after confirming what specific information had been accessed.

This incident has surfaced amid increasing concerns over artificial intelligence and its implications. Both Altman and Anthropic CEO Dario Amodei recently alerted the UN Security Council about potential threats posed by rapidly evolving AI technologies. Altman stressed, “There’s a risk that we could lose control of our future to AI,” highlighting the need for vigilant oversight to ensure governments and industries keep it in check.