Mandatory AI guardrails for high-risk AI (in development)
Australian Mandatory Guardrails for High Risk AI Settings — Treasury consultation in 2024/2025.
Who must comply
Developers + deployers of 'high-risk' AI systems (definition TBD).
What triggers it
Final regulations + commencement.
When due
TBD; expected 2026-2027.
Evidence required
AI risk register, model documentation, oversight + accountability records.
Max penalty
Expected to mirror EU AI Act-style civil penalties (substantial)
Summary
DISR's Mandatory Guardrails for High Risk AI Settings (proposals paper 2024) follows the EU AI Act pattern. Expected to require for 'high-risk' AI uses: risk management, data governance, transparency, human oversight, accuracy + robustness, accountability. Commencement TBD (likely 2026-2027).
Industries
Topics
Source: https://www.industry.gov.au/publications/introducing-mandatory-guardrails-ai-high-risk-settings-proposals-paper. Rules Mate is not a law firm. Always verify against the live regulator source before acting.