§30 · Lane 7 — Australian Regulatory Primary Instruments

DISR Mandatory Guardrails (Proposals) the AU comparator to EU high-risk AI

DISR (2024) · DISR Guardrails

Regulatory Tier 2 Lane 7 Stable URL
Read on publisher · Stable URL

Bibliographic data

Title
DISR — Introducing mandatory guardrails for AI in high-risk settings (Proposals paper, September 2024)
Authors / Issuing body
Department of Industry, Science and Resources (DISR)
Venue / Publisher
Commonwealth of Australia
Year
2024
Designation
Regulatory
Licence
Stable URL — refer to publisher for full licence terms.

How to cite

DISR (2024). DISR — Introducing mandatory guardrails for AI in high-risk settings (Proposals paper, September 2024). Commonwealth of Australia. https://consult.industry.gov.au/ai-mandatory-guardrails.

The Commonwealth proposals paper on regulating high-risk AI through mandatory guardrails, proposed approaches to defining high-risk AI, and three regulatory options for mandating the guardrails (option 1 sector-by-sector; option 2 framework legislation; option 3 dedicated AI Act). The Australian Government has indicated it will not proceed with the proposed mandatory guardrails approach; feedback fed into the National AI Plan.

Why it matters for NETEVO

The September 2024 DISR proposals paper remains significant even though the Australian Government has indicated the mandatory-guardrails approach will not proceed. The paper set out proposed approaches to defining high-risk AI and three regulatory options for mandating the guardrails: option 1, a sector-by-sector approach; option 2, framework legislation; option 3, a dedicated AI Act. The consultation feedback fed into the National AI Plan, which now signals the current policy direction — but the proposals paper itself remains the formal regulator articulation of high-risk AI in Australian terms.

The AU high-risk-AI vocabulary anchor. The proposals paper is the canonical AU regulator statement of what high-risk AI would have looked like under an Australian regulatory regime. It is the AU comparator to the EU AI Act high-risk categorisation, expressed in Australian rather than imported European vocabulary, and it remains the most direct domestic reference for defining high-risk AI in an Australian context.

A substantive policy posture shift. The decision not to proceed with mandatory guardrails — and the redirection of feedback into the National AI Plan — is itself a material data point. Organisations that prepared for a mandatory-guardrails regime are now operating against a different signalled trajectory, and the September 2024 consultation snapshot should be read alongside the National AI Plan rather than as current policy.

Paired-by-design with the voluntary standard. The ten proposed mandatory guardrails substantially overlap with the ten voluntary guardrails in the DISR Voluntary AI Safety Standard. The two instruments were designed as a pair: voluntary today, mandatory for high-risk settings tomorrow. Even with the mandatory limb on hold, the guardrail design itself remains the stable AU reference, and demonstrable evidence against those guardrails is what a regulator would be positioned to inspect under any future regime.

Where NETEVO applies this

Related audiences