Australia Moves to Mandate Independent Audits of Frontier AI Models
The Australian government plans binding safety rules for leading AI companies, including mandatory independent model audits, a new regulatory authority and heavy penalties for violations.
The Australian government under Prime Minister Anthony Albanese wants to legally require leading AI developers to subject their so-called frontier models to independent audits. The proposal provides for binding safety rules that would be enshrined in law for the first time. Australia would thus take a step that only a few countries have taken so far.
According to the available information, a newly created regulatory authority would enforce compliance with the rules. Independent audits of the AI models would be mandatory. For violations, significant penalties are reportedly planned. The exact amount of the penalties, the legal text and a timetable for the legislative process have not yet been published.
The government in Canberra's set priorities are noteworthy. The regulation is to be implemented initially at the national level, while international coordination takes a back seat for now. How the term "frontier models" is to be defined in the law is not described in detail in the information available so far. It therefore remains unclear which systems and companies would be specifically affected.
The initiative comes at a time when governments worldwide are debating how powerful AI systems can be regulated without unnecessarily slowing innovation. Whether the Australian model will hold up in this form now depends on the further political process.
Source: the-decoder.com
17
