Australia announces AI regulator, considers criminal law changes

The Australian government has announced a new AI safety regulator to enforce penalties against companies that breach obligations under its planned regulatory framework. Michelle Grattan reports in The Conversation that Prime Minister Anthony Albanese outlined the measures on November 10. The proposals target companies developing the largest and most advanced AI models. The announced duties form part of proposed legislation and are not presented as rules already in force.

According to The Conversation, the proposed framework would impose six duties: maintain security controls throughout development and deployment; identify and manage serious model risks; promptly notify the Australian government of incidents; allow independent evaluations; maintain and publish safety and security frameworks for governance, risk management and transparency; and adapt to evolving standards and rules. The report does not specify a commencement timetable.

Grattan also reports recommendations from a government review calling for exploration of criminal penalties for particularly serious regulatory non-compliance and/or a separate failure-to-prevent offence. The latter would concern a company responsible for developing an AI model that engages in harmful conduct, where the company fails to take reasonable steps to prevent that harm. These are options for consideration, rather than offences already introduced. According to the report, Albanese said the government had accepted all 18 recommendations and begun work on them. This account rests on secondary reporting, without direct verification of the legislative text or the full review report.