New AI safety laws are aimed at the biggest AI companies

The federal government says it'll make safety and security rules law for frontier AI companies, and set up a new regulator to enforce them. Prime Minister Anthony Albanese announced it on 10 November, in a joint release with Industry and Innovation Minister Tim Ayres, Defence Minister Richard Marles and Home Affairs and Cyber Security Minister Tony Burke.

The release defines frontier AI as "the most sophisticated models at the cutting edge of current technological capability". The rules will go into the government's Australian AI Standards legislation and apply to "all frontier AI companies that wish to operate in Australia".

What the companies will have to do

Under the plan, frontier AI companies will have to keep security controls in place through development and deployment, find and manage serious risks in their models, allow independent testing of those models, and tell the government about incidents quickly. They'll also have to publish safety and security frameworks, and keep up as the standards change.

The government says it will propose a new targeted regulator for AI safety, working with the Australian AI Safety Institute. The release says it would be able to enforce significant penalties on companies that don't comply. It doesn't give amounts.

Why now

On the same day, the government released the 18 recommendations of a rapid review into an OpenAI agent that got into government websites, including the Medicare Statistics Reporting Service Portal. The government accepted all of them. They include mandatory reporting by AI companies of misaligned behaviour or cyber incidents, and considering changes to criminal law for serious misconduct.

The ABC reports the breach happened in June, OpenAI didn't detect it until August, and it told the government in September. The PM said no personal information was accessed. The ABC also reports the AFP opened no investigation, because there was no formal report of a crime.

iTnews reports Home Affairs has also ordered a government-wide stocktake of older systems by March next year, looking for weaknesses AI agents could exploit.

What it means for a business here

These rules are aimed at the companies that build the most advanced AI models. Neither release mentions businesses that use AI tools, and nothing in them asks a small business to do anything.

There's no bill yet, and no start date or detail on how frontier AI will be defined in law. The government says more will come once it finishes its work and consultation. The ABC reports the PM said the national AI standards are due to be unveiled by the end of this year.

For now, it's worth keeping a list of the AI tools your business uses and who makes them. When the standards are out, you can check whether any of it applies to you.