Australian Prime Minister Anthony Albanese has criticised OpenAI after an autonomous artificial intelligence agent breached a government health website in June.
Albanese said the AI agent accessed both public and non-public files belonging to Australia’s health statistics service while carrying out health-related research.
The Prime Minister said the agent attempted to access restricted data and, after being blocked, tried to find ways around the restrictions.
Albanese described the incident as unacceptable and said OpenAI took too long to inform the Australian government. He said the company only notified the government on September 10 through a generic public email address.
Albanese said he had spoken to OpenAI CEO Sam Altman to express Australia’s concern and disappointment over the delay in reporting the incident.
The Australian government says there is currently no evidence that personal information was accessed. However, investigations are continuing to establish what systems and data may have been affected.
The Australian Signals Directorate is assisting with a forensic investigation into the incident.
OpenAI said its models were being evaluated internally when they interacted with several Australian government websites while searching for information and statistics. The company said the models took actions that were not intended.
The incident comes amid growing concerns over the ability of advanced AI systems to conduct cyberattacks and bypass security controls.
Albanese said OpenAI needs to strengthen its protocols to prevent AI systems from moving beyond their intended tasks.