Policy Briefing — Australia Issues Interim Response on Safe and Responsible AI

Executive briefing: On 1 July 2024 the Australian Government released its Interim Response to Safe and Responsible AI in Australia. The paper commits to legislating mandatory guardrails for high-risk AI systems, establishing a voluntary AI Safety Standard and watermarking guidance, and empowering regulators such as the Australian Human Rights Commission and the eSafety Commissioner to address AI harms.

Key commitments

High-risk guardrails. The government will develop binding requirements for AI used in critical infrastructure, essential services, employment, education, and biometric surveillance, including mandatory risk assessments and record-keeping.
AI Safety Standard. Standards Australia will lead a voluntary standard focused on safety testing, transparency, and responsible release of general-purpose AI.
Watermarking guidance. The interim response tasks the Digital Transformation Agency with publishing best practices for watermarking and labelling synthetic content.
Regulator coordination. A new AI Taskforce will coordinate agencies including ACMA, ACCC, AHRC, OAIC, and eSafety on enforcement, guidance, and sandboxes.
Support for SMEs. The government plans tailored guidance and funding programmes to help small businesses adopt AI safely.

Next steps

Targeted consultation. Throughout 2024 the Department of Industry, Science and Resources will consult on the design of high-risk guardrails.
Legislation development. Draft legislation is scheduled for release in 2025 following consultation outcomes.
Standardisation work. Standards Australia will commence drafting the AI Safety Standard in partnership with industry and academia.

Program actions

Risk scoping. Identify AI use cases in Australia that fall within high-risk sectors and start documenting risk assessments, testing evidence, and human oversight plans.
Transparency tooling. Prepare to implement content provenance and labelling workflows aligned with forthcoming watermarking guidance.
Regulator engagement. Track consultations from OAIC, eSafety, and ACCC to align compliance strategies with sector-specific expectations.
Standards mapping. Monitor development of the AI Safety Standard and align control frameworks with ISO/IEC 42001 and NIST AI RMF for reuse.

Sources

Zeph Tech is partnering with Australian organisations to operationalise emerging AI guardrails, from risk testing to provenance controls.