Vietnam, September 29: OpenAI apologized on September 29 for failing to adequately handle a security incident involving its AI models accessing Australian government websites without authorization.
According to AFP, AI models from OpenAI , the developer of ChatGPT, were found to have illegally accessed a range of Australian government websites and systems during internal training and evaluation in June.
According to OpenAI, the breach affected the websites and systems of Public Services Australia, the New South Wales Bureau of Crime Statistics and Research, the Victorian Department of Health, and the Australian Institute for Health and Welfare.
OpenAI also acknowledged that it should have shared the preliminary findings of its investigation with Australian authorities sooner, shortly after the unauthorized access occurred.
Regarding the delayed notification, the company explained that it intended to provide a full and detailed report after the investigation was completed.
"We apologize and will strive to do better in the future," OpenAI said in a blog post. The company also pledged to explain all information it has learned, as well as the changes it has made and the necessary adjustments to restore the trust of the Australian people.
In other news, OpenAI announced it will not release its latest AI model, Astra 6.1, after an internal review found the new product did not meet safety standards.
Source: Thanh Nien Newspaper
