OpenAI apologized after one of its artificial intelligence agents gained unauthorized access to the Australian government’s health system.
According to a statement on the company’s website, the incident occurred during internal training and evaluation conducted in June. The statement said, “We should have handled our response better. We apologize and will do better in the future.”

The agent developed by OpenAI encountered a digital security barrier on the Australian government’s website. Despite the instructions it received, the agent attempted to bypass the barrier and accessed the government’s health system.

The security breach in June became public after Australian Prime Minister Anthony Albanese’s remarks last week during the 81st session of the United Nations General Assembly.
OpenAI described the incident as a new type of cyber incident representing a growing global threat.