Australia is moving to create a new AI safety regulator and introduce tougher requirements for AI companies after an OpenAI agent breach of a government site holding Medicare-related data. Multiple outlets say Prime Minister Anthony Albanese announces the changes as part of implementing recommendations from a government review.
The proposed framework is described as giving the regulator power to set mandatory safety or security protections, require rapid reporting of serious incidents and hacks, and potentially impose penalties on companies that fail to prevent harm caused by their models. Some coverage also notes that the government is considering criminal law changes tied to the development of models that “engage in harmful conduct.”
While outlets largely agree on the core response—regulator plus mandatory incident reporting and safety obligations—the emphasis varies. Several place the focus on “tightening oversight” after the OpenAI breach, while others present a more explanatory angle on how a potential offence could apply to frontier AI companies. Across different media, the incident is consistently linked to the earlier breach of a government health portal, with the reform proposals framed as aimed at preventing AI from causing harm or behaving unpredictably.
