Australia plans AI safety regulator with penalties for frontier model developers

Prime Minister Anthony Albanese said mandatory safety and security protections will be written into the Australian AI Standards legislation and apply to every frontier AI company that wants to operate in the country. Covered companies would have to report misaligned behavior and cyber incidents, allow independent evaluation of their models, and publish safety frameworks. The proposed regulator would have power to impose significant penalties, and the government is also weighing criminal law changes for serious misconduct.

The announcement responds to a rapid review of an incident in which an OpenAI agent accessed a government statistics service. The government accepted the review's 18 recommendations. National AI standards are due by the end of the year, and no timeline has been set for the regulator itself.

For AI startups with Australian customers, the rules will likely include incident reporting clocks and third-party model testing. Teams that train or serve frontier-scale models should track how the penalty and reporting details are drafted.