Australia unveils AI standards to govern the deployment of technology

No coercive law nor laissez-faire: Canberra chooses the path of standards to guide adoption.

In plain terms

Australia has presented a set of national standards aimed at guiding the deployment of AI - a framework for guidance rather than binding legislation like the European AI Act.

Our analysis (follow-up - ai-governance-standards)

Three approaches coexist in the world: the hard rule (the European AI Act), the assumed laissez-faire, and the path of standards. Canberra chooses the third. The advantage: flexibility, faster adoption, no risk of freezing a moving technology in law. The disadvantage: standards without teeth do not constrain anyone; their real scope depends on the voluntary adherence of companies and future public markets that would make them de facto mandatory. This is the bet of "soft law": guiding the market before constraining it - similar to Demis Hassabis' proposal for an independent body of standards for the frontier. For a medium-sized country, it is also a way to exist in the global governance of AI without the regulatory means of a large bloc.

To watch

The precise content of the standards (safety, transparency, bias), their articulation with the AI Act and the ISO/IEC 42001 standard, and above all whether they become a condition for access to public procurement - the only lever that transforms a voluntary standard into a real obligation.