Mandatory guardrails for AI in high-risk settings
Binding rules for AI used in high-risk settings, covering testing, transparency and accountability. The Government consulted on them in 2024 and did not adopt them.
What exists now
From 5 September to 4 October 2024 the Department of Industry, Science and Resources consulted on proposed mandatory guardrails for AI in high-risk settings. The National AI Plan, published on 2 December 2025, did not adopt them. It relies on existing privacy, consumer, online safety, anti-discrimination and critical-infrastructure law, and on voluntary guidance for industry, the Guidance for AI Adoption.
Where it stands
Nothing here yet.
Risks this could address
Who is calling for this
Good Ancestors Policy
Introducing mandatory guardrails for AI in high-risk settings, Position paper Read it ↗
“Good Ancestors recommends that Australia develop a set of mandatory guardrails for the development of GPAI with systemic risk that draws on international models and the best available research.”
Shop, Distributive and Allied Employees' Association (SDA)
Submission 17 to the Joint Select Committee on Artificial Intelligence (2026), Recommendation 12 Read it ↗
“The ten voluntary guardrails set out in the Commonwealth's Guidance for AI Adoption should be made mandatory for AI systems that materially affect employment, safety or consumer welfare”
Victorian Trades Hall Council (VTHC)
Submission 293 to the Joint Select Committee on Artificial Intelligence (2026), Recommendation 6 Read it ↗
“Recommendation 6: Subject high-risk AI systems to mandatory oversights and controls and classify AI systems used for employment related purposes as high-risk.”
Workday
Submission 51 to the Joint Select Committee on Artificial Intelligence (2026) Read it ↗
“We support the current voluntary standards approach for general AI use, paired with tailored mandatory safeguards for high-risk use cases.”
ARC Centre of Excellence for Automated Decision-Making and Society
Submission 203 to the Joint Select Committee on Artificial Intelligence (2026) Read it ↗
“The government should reconsider the past proposed mandatory guardrails for AI, including basic documenta>on, risk management, tes>ng and monitoring obliga>ons for high-impact applica>ons of AI”