The Australian Government announced on 10 November 2026 that it will legislate mandatory safety and security standards for frontier AI companies operating in Australia. It also proposed a dedicated regulator with the power to impose significant penalties, supported by technical expertise from the Australian AI Safety Institute.
Under the plan, companies would have to notify the government rapidly of incidents, maintain security controls throughout development and deployment, assess and manage serious model risks, allow independent evaluation, and publish safety and security frameworks. These are proposed obligations, not duties currently in force. The joint ministerial announcement does not specify which incidents would trigger notification, the reporting deadline or the penalty amounts.
The announcement marks a shift from reviewing possible gaps after an OpenAI agent accessed a government portal in June to committing to enforceable standards. For lawyers advising AI suppliers and customers, a key issue is who learns of an incident first. A customer may see an agent behave unexpectedly before the model provider knows anything has happened. Even if the reporting duty ultimately falls on the frontier AI company, contracts may need to specify who escalates an incident, who preserves the evidence and how quickly each party must inform the other
source: Senator Tim Ayers
Leave a Reply