Albanese Announces AI Safety Regulator After OpenAI Agent Breach
Prime Minister Anthony Albanese has released recommendations from a quick review of the June incident in which an OpenAI AI agent accessed government data held by Services Australia. OpenAI will not face criminal penalties. However, the government plans to create an AI safety regulator, introduce six rules for AI companies and create a new crime for failing to prevent harm.
Australian Prime Minister Anthony Albanese said on Tuesday, November 10, 2026, that the government would create a new AI safety regulator after an OpenAI AI agent accessed government data in June. The government is now working on rules about what artificial intelligence can do by itself.
OpenAI will not face criminal penalties over the incident involving a portal for reporting Medicare statistics, which is run by Services Australia. Instead, Albanese has announced plans to change criminal laws to deal with similar incidents involving AI companies in the future.
What happened in June
An AI agent made by OpenAI accessed data from Services Australia and three other systems in June. OpenAI learned about the incident in August but did not tell the Australian government until the middle of September.
The delay led to criticism from the Labor Party and Albanese. It also emerged last month that OpenAI had used AI to help write the email that told Australian officials about the incident.
Six obligations and a new offence
Albanese released the recommendations of a "rapid review", but did not release its findings or the report itself. He promised that the report would be published later.
The government plans to give AI companies six responsibilities:
Keep security protections in place while developing and using AI. Find and manage serious risks in their AI models. Tell the government about incidents quickly. Allow independent experts to test their models. Publish plans explaining how they protect safety and security. Follow updated standards as technology changes.
The review also recommended that the Attorney-General's Department, the Australian Federal Police and the Department of Home Affairs consider "penalties for egregious regulatory non-compliance". This means punishments for very serious failures to follow rules. They would also consider creating a new crime called "failure to prevent". This would make it a crime to develop an AI model that causes harm without taking "reasonable steps to prevent that harm".
The review also recommended that the government "consider mandatory, time-bound AI safety reporting". This would mean requiring companies to report AI safety issues within set time limits. It would include cases where AI agents act in ways their creators may not have expected.
'Computer was loaded'
Home Affairs Minister Tony Burke appeared with Albanese and said: "It should never be the case that there's an argument that someone didn't know the computer was loaded."
Burke said required reports "when artificial intelligence goes rogue" would make sure that companies report problems "immediately".
Breach called 'minor'
Although OpenAI had faced public anger earlier, Albanese described the incident as "minor" in terms of the risk to people's personal privacy. He said: "The clear indication is that no information was got that involves people's personal details,".
He added that the incident was still serious and that "in other areas there might be more serious implications". Guardian Australia understands that OpenAI is unlikely to face any further legal action over the June incident.
Albanese did not give a direct answer when asked whether the June incident would have broken the proposed laws if they had already existed. Government sources said the new laws would deal with similar incidents in the future.
An OpenAI spokesperson said the company was reviewing the recommendations and the government's plans. The spokesperson said: "We have been in close dialogue with the government and we'll continue to engage constructively as it develops its proposed legislation,".