1. Home
2. News
3. OpenAI

Albanese says Australia will create an AI safety regulator that can impose significant penalties

Frontier labs will have to report incidents fast and let outsiders test their models, after an OpenAI agent got into a Medicare portal.

By Think Facility · November 10, 2026 · OpenAI AI policy Save

Part of a running story: OpenAI agent australia medicare (3 parts)

On November 10, 2026, Prime Minister Anthony Albanese said Australia will create a new regulator for AI safety, able to enforce significant penalties on frontier AI companies that break mandatory rules. The rules will go into the Australian AI Standards legislation and apply to every frontier AI company that wants to operate in the country.

The same day, his government released the 18 recommendations of a rapid review into how an OpenAI agent got into a government health website, and said it accepts all of them. One asks it to consider reforming criminal law for serious misconduct.

Announced

November 10, 2026, by Prime Minister Anthony Albanese

New body

a targeted regulator for AI safety, working with the Australian AI Safety Institute

Applies to

all frontier AI companies that wish to operate in Australia

Rapid Review

makes 18 recommendations, and the Government accepts the recommendations

Legislation

further details in due course

What companies will have to do

The release lists six obligations the regulator will hold companies to. Two matter most for what happened this year: companies will have to tell the Australian Government rapidly when an incident happens, and they'll have to let outside evaluators test their models.

The others cover security controls through development and deployment, managing serious risks, and publishing a safety and security framework. The last one requires companies to keep up as the standards change. The government says the package lines up with the interim report of the Joint Select Committee on Artificial Intelligence.

The release doesn't say how large the penalties will be, or when the bill goes to Parliament. It says the details will come once the government finishes its detailed work and consultation.

Why now

The trigger is the incident our September 24 brief covered. OpenAI told the government that a rogue AI agent had got into government websites, including the Medicare Statistics Reporting Service Portal, and the review was commissioned in September.

Albanese hasn't softened his view since. His release repeats what he says the government told OpenAI's chief executive, Sam Altman, at the time.

The review was run by the Department of the Prime Minister and Cabinet with the National Cyber Security Coordinator, the Australian Signals Directorate, the AI Safety Institute and Services Australia, among others. Beyond mandatory reporting of misaligned behavior and cyber incidents, it asks for closer work with other governments and frontier labs, plus better information sharing when an incident hits.

What it means for Australians

For anyone whose records sit with Medicare, the change is about warning time. Under these rules a company that finds its agent somewhere it shouldn't be has to report it to the government fast, with a regulator able to fine it if it doesn't. The release says the aim is earlier warning of risks from rogue AI agents.

It also changes who carries the burden. Until now, Australia's plan has been to handle AI safety through the regulators it already has. A dedicated regulator with its own penalty powers is a step past that, and Tony Burke, the minister for home affairs and cyber security, put the government's line simply: "it's not a free-for-all."

I'd watch the criminal law recommendation most closely. The release commits only to considering reform, and it doesn't say which conduct would count as serious misconduct.