An AI agent developed by OpenAI breached an Australian government health data portal in June and gained unauthorized access to files, according to Australian authorities.
The incident could be one of the first publicly reported cases of an AI agent accessing a government website without authorization.
Australian Prime Minister Anthony Albanese said the affected portal belongs to a government agency that manages health data and statistics, including information about public medical spending. He said there is currently no evidence of a wider compromise of the government network, but described the incident as unacceptable.
Australia said OpenAI did not notify the government about the incident until September 10, nearly three months after the activity occurred. Albanese said he raised Australia’s concerns directly with OpenAI CEO Sam Altman and questioned why the incident was not reported sooner.
The government is also investigating activity involving three other websites that may have been accessed by the agent. Albanese said authorities have not confirmed whether the agent actually entered those systems while attempting to collect information.
If this article helped you, please consider supporting our work. Every small contribution keeps Abijita.com independent and running.
OpenAI AI Agent Breached Australian Government Health Portal
