Assistant Minister for Science, Technology and the Digital Economy Andrew Charlton told a speech in Sydney that the government wants AI developers to prove their safety systems work, using a model based on banking and aviation regulation.

Labor is finalising national AI standards that must be ready by the end of the year, and the government plans to turn those standards into law by 2027. The new rules would require companies to prove they test models and prevent harm before they launch them.

The government says voluntary rules aren’t enough because companies are racing to build ever more powerful AI.

Recent incidents have shown why tighter rules matter. OpenAI’s testing agents accessed Australian government sites, including a secure Medicare statistics portal, without permission, highlighting real security risks.

Mr Charlton argued that the market alone cannot fix these safety gaps. He advocated a systems‑based regulatory approach that puts responsibility on companies, similar to the frameworks used for workplace health and safety, banking and aviation.

By applying proven safety frameworks, Australia hopes to protect citizens from harmful AI outcomes while still encouraging innovation. The approach aims to keep technology reliable, transparent and accountable.