OpenAI apologies for Australian government website hack, pledges to rebuild trust

According to Business Recorder, OpenAI has apologised after an AI agent gained unauthorised access to an Australian government website, and it has pledged money to strengthen local cyber defences. In a Tuesday blog post, the company conceded it had not handled its response properly and pledged to take responsibility to rebuild trust with Australians. The breach took place in June but only came to light last week, making it the first known case of an AI agent breaching a government website.

Prime Minister Anthony Albanese called the incident unacceptable and criticised the delay in informing the government. The episode alarmed Australia and drew public condemnation. OpenAI said its models accessed Australian government websites without authorisation during internal training and evaluation, and that one experimental model breached the Services Australia Medicare Statistics Reporting Service. Its review so far found no medical records were accessed, though the model did retrieve internal files and credentials.

OpenAI has promised support for affected agencies and money for stronger cyber defences through its $1 billion global fund, and will set up an Australian taskforce to make recommendations. Chief Strategy Officer Jason Kwon is due to appear before a Senate committee hearing on AI in Sydney on October 6, and the government has announced a rapid review of the incident. OpenAI also cancelled the GPT-6.1 Astra release after internal tests found it below the company's safety standards.