OpenAI chief strategy officer Jason Kwon apologised for its AI models breaching Australian government websites at a parliamentary hearing in Sydney on 6 October and pledged faster responses to similar incidents, Bloomberg reported. The pledge follows June access to federal health data and New South Wales bushfire statistics, with Australian officials notified of the incidents in September and October, ABC News reported.
Key points
- Kwon told the parliamentary inquiry that the activity should not have happened and OpenAI should have responded better.
- OpenAI says it will work to prevent similar incidents and respond faster if they occur.
- Prime Minister Anthony Albanese said the company took too long to notify the government about access to Medicare data.
Kwon apologises to the inquiry in Sydney
“I want to begin with an apology,” Kwon told Australia’s Joint Select Committee on Artificial Intelligence. He said the activity should not have happened and that OpenAI should have handled its response better.
“We are sorry, and we know we have work to do to rebuild trust with the Australian people,” Kwon said. OpenAI pledged to try to stop a repeat and act more quickly should another breach occur.
The apology follows the disclosures that brought the company before lawmakers. AI Affairs previously reported that the inquiry would question OpenAI over access to Medicare data. Kwon’s commitment on 6 October addresses how the company will handle another incident.
Medicare portal access drew Albanese’s criticism
An OpenAI agent accessed public and non-public files associated with the Australian Medicare Statistics Reporting Portal in June, Prime Minister Anthony Albanese told reporters. OpenAI said it discovered the activity in August during a review of what it called “misaligned model activity” and notified Australian officials on 10 September, according to ABC News.
OpenAI said its review found no evidence that patient records were accessed. It said the material included aggregate health statistics and internal file names, and that its models took unintended actions while looking for answers and statistics about Australia during an internal evaluation.
Albanese said he spoke with OpenAI chief executive Sam Altman to express Australia’s “extreme concern”. He also said “it took the company way too long to inform the government” and criticised the way it delivered that notification.
NSW bushfire data was accessed in June
OpenAI also said one of its models queried the NSW National Parks and Wildlife Service’s Fire History service and gathered summary bushfire statistics that were not publicly available. The NSW Premier’s Department said it believed the incident occurred in June and that OpenAI informed officials on 1 October.
OpenAI became aware of the NSW breach on 29 September and spent 48 hours reviewing its scope before informing the premier’s office, The Guardian reported. OpenAI said the results it reviewed did not show the model retrieving personal information.
The NSW Department of Climate Change, Energy, the Environment and Water is working with the state’s cyber security agency to investigate the breach, The Guardian reported. The Australian Signals Directorate has also been informed.
