OpenAI agent broke into Australian government website, Prime Minister openly condemned

Author, Harry Sekulich

and

Author, Lana Lam

Designation, Sydney

Published 2 minutes ago

Reading Time: 6 minutes

Experts say that in June, an unauthorized agent of OpenAI had hacked an Australian government website. AI had gained access to personal data.

This is the first case of its kind in the world.

Australian Prime Minister Anthony Albanese said in New York on Wednesday that an OpenAI agent had "infiltrated" a portal containing "non-sensitive" data related to Australia's universal health service scheme Medicare.

He openly discussed with OpenAI CEO Sam Altman the "extreme delay" in making the data leak public, saying there would be "legal consequences".

OpenAI said it became aware of the breach in August when it was reviewing "misdirected model activity." After this, on September 10, he sent this information to the general email inbox of an Australian government agency.

Five days later, the government agency Services Australia forwarded the email to Australia's Cyber ​​Security Centre. After this, a government minister was informed about this and then the Prime Minister was informed.

Albanese said he spoke to Altman and expressed Australia's "deep concern" about the incident and his "disappointment" at the company's months-long delay in reporting the breach. He also said that he is disappointed with the way the company gave this information.

The Australian Prime Minister said Altman acknowledged there were 'some protocol issues' with OpenAI.

Skip to Most Read Skip to content

most read

finished

PM talked about 'forensic investigation'

skip ahead podcast

Dinbhar: Same voice, new style

Big news from the country and the world that made headlines.

episode

finished

Albanese said a 'forensic investigation' would be led by the country's cyber security agency to determine whether other government systems were also affected.

He said that in this investigation it will also be seen whether there is a need to send this matter to the police. He said there would be 'obviously legal consequences'.

Albanese elaborated on the breach, saying it involved "public and non-public files" of the Medicare Statistics Reporting Service portal. This portal contains "non-sensitive" data and statistics.

Three other systems of the government may also be affected. These include the Australian Institute of Health and Welfare and two state agencies, the New South Wales Bureau of Crime Statistics and Research and Victoria's Department of Health.

Albanese said, "At this time it is not believed that anyone's personal information has been accessed, but the investigation is ongoing."

“Nevertheless, this situation is clearly unacceptable,” he said.

What did OpenAI say in clarification?

OpenAI said in a statement that it "identified activity by our models on a number of Australian government websites and services while exploring available data and answering questions about Australia during an internal assessment."

"During this time, our models did things that we had not intended," the statement said.

Albanese declined to answer questions about whether he had raised the issue during a face-to-face meeting with US President Donald Trump in New York on Tuesday night. World leaders have gathered in New York for the United Nations General Assembly.

Australia was among 22 countries that signed a joint statement earlier this week calling for global monitoring and safeguards for AI development.

Cyber ​​security experts told the BBC that the incident was a warning to regulators as AI agents become more readily available for personal and commercial use.

Dr Hammond Pearce, a senior lecturer at the Institute for Cyber ​​Security at the University of New South Wales, told the BBC that this was the first known case in which AI agents decided to break into a government institution of their own free will, but that more such cases would emerge.

He said, "I hope such attacks will continue to happen." He said these attacks were "likely to increase in both severity and number."

He said, "I hope this incident will start ringing alarm bells for governments around the world."

What is the debate about AI?

Earlier this year, OpenAI reported that a group of AI agents it was testing had gone out of control and had secretly tried to hack another tech company called Hugging Face.

Many other such incidents related to AI have also come to light this year. These include a case in which a digital assistant removed a person from the waiting list for a Pilates class without any instructions, so that an Australian man could get a place in that class.

Many prominent people of AI companies, including Altman, Dario Amodei of Anthropic and Elon Musk have themselves said that the speed with which AI is developing is dangerous for humanity and its speed needs to be controlled.

But America and China, which are competing to achieve supremacy in AI, are big obstacles in its path. Both countries are not in favor of much regulation. They want to reap the economic and technological benefits of AI and downplay security concerns.

We used AI to translate this article, which was originally written in English. A BBC journalist checked this translation before publication. Learn more about how we're using AI.