AI safety policy ideas experts have proposed

Researchers, engineers and governance specialists in the AI safety community have put forward a number of ideas for making the development of advanced AI safer. Below we explain what some of the most-discussed ideas involve, and why the experts who proposed them believe they would help.

This page is background information to help you understand the AI safety policy conversation. The ideas below are recommendations made by independent experts; we share them so the public can understand the options being discussed. It is educational analysis, not a call to support any particular law or to take any specific action.

AI Safety Institute

A well-resourced, independent technical body to assess AI risks and advise on safety standards.

Only $1 is spent on ensuring AI systems are safe for every $250 spent making them more powerful. A well-funded Australian AI Safety Institute would help address this imbalance.

Learn more

Mandatory Guardrails

A dedicated AI Act with mandatory guardrails for high-risk AI systems will both protect Australians and create the certainty businesses need to innovate.

Clear safety standards for high-risk AI systems are essential to protect against serious harms while credibly building the public trust needed to unlock AI's estimated $600 billion economic potential for Australia.

Learn more