Australia will establish an artificial intelligence safety regulator with the power to penalize non-compliant tech companies. Frontier developers must continuously test systems and prevent harms, with binding legislation planned for 2027.