Voluntary AI Safety Standard (10 Guardrails)¶

Purpose: Detailed control catalogue for implementing Australia's 10 AI safety guardrails Audience: Governance, risk, compliance and technical teams | Time: 45-60 minutes for full review

Australia's Voluntary AI Safety Standard (VAISS) provides 10 practical guardrails organisations can adopt to develop, deploy and use AI safely and responsibly.

The National AI Centre, within the Department of Industry, Science and Resources, published VAISS on 5 September 2024. The Department updated the online publication in December 2025 to point readers to the newer Guidance for AI Adoption.

The standard is voluntary and does not create new legal duties. It complements existing Australian law by helping organisations identify risks, controls and relevant obligations.

Importantly, the 10 guardrails are consistent with leading international standards and frameworks, including:

Current status

On 21 October 2025 the National AI Centre released the Guidance for AI Adoption, which sets out 6 essential practices (AI6) for responsible AI governance and adoption. The Department describes it as updated and simplified guidance that evolves VAISS.

The AI6 implementation guidance includes an appendix that maps the original VAISS controls to the new practices. The 10 guardrails are best used as a detailed control set and historical reference, especially where contracts, risk registers or external frameworks still refer to VAISS.

A consultation process for a proposed VAISS v2 ran in early 2025. As at 22 July 2026, the Government's publication catalogue does not contain a standalone VAISS v2. Use the published AI6 guidance for current implementation work and the original VAISS guardrails where a detailed or legacy mapping is useful.

Australian Standards for AI framework announced — July 2026

On 15 July 2026, the Government announced plans to legislate a framework for large data centres and AI training, including future mandatory requirements for large AI data centres. The announcement does not withdraw or make the VAISS guardrails mandatory. Organisations can continue using these guardrails and AI6 as non-binding guidance while complying with existing law. (Prime Minister of Australia; PM&C Office of AI, accessed 22 July 2026)

Why this matters¶

Adopting the guardrails early helps organisations build trust, resilience and regulatory readiness. By embedding these practices now, businesses can:

🛡️ Reduce risks from bias, errors and misuse of AI
🤝 Strengthen transparency and customer confidence
🚀 Build evidence of systematic AI governance and risk management
⭐ Demonstrate leadership in responsible AI adoption

The 10 guardrails¶

Establish, implement and publish an accountability process, including governance, internal capability and a strategy for regulatory compliance.
Establish and implement a risk management process to identify and mitigate risks.
Protect AI systems, and implement data governance measures to manage data quality and provenance.
Test AI models and systems to evaluate model performance and monitor the system once deployed.
Enable human control or intervention in an AI system to achieve meaningful human oversight across the life cycle.
Inform end users about AI-enabled decisions, interactions with AI and AI-generated content.
Establish processes for people impacted by AI systems to challenge use or outcomes.
Be transparent with other organisations across the AI supply chain about data, models and systems so they can address risks.
Keep and maintain records to allow third parties to assess compliance with the guardrails.
Engage stakeholders and evaluate their needs and circumstances, with a focus on safety, diversity, inclusion and fairness.

What the guardrails do

✅ Encourage transparency and accountability for AI systems
✅ Require risk assessment, testing and human oversight before and after deployment
✅ Promote record-keeping and supplier due-diligence across the AI supply chain
✅ Emphasise stakeholder engagement and ongoing monitoring as systems evolve

How this maps to the 6 essential practices (AI6)¶

The Guidance for AI Adoption condenses the 10 VAISS guardrails into 6 essential practices. The table below is a practical SafeAI-Aus summary, not a substitute for the detailed crosswalk in the official AI6 implementation guidance.

AI6 practice	Closest VAISS guardrails and themes
Decide who is accountable	1. Accountability; 9. Records
Understand impacts and plan accordingly	2. Risk management; 7. Challenge processes; 10. Stakeholder engagement
Measure and manage risks	2. Risk management; 3. System and data protection
Share essential information	6. End-user information; 8. Supply-chain transparency; 9. Records
Test and monitor	3. System and data protection; 4. Testing and monitoring
Maintain human control	5. Human control; 7. Challenge processes

How to use this mapping

📊 Use AI6 as your top-level framework when explaining AI governance to boards, executives and regulators
📋 Use the 10 guardrails in this page as a control library when:
- building or updating AI policies and standards
- designing AI risk assessments and risk registers
- writing contractual clauses and supplier questionnaires
🔗 Check the official AI6 crosswalk before claiming exact equivalence between a VAISS control and an AI6 action

How to use this in your business

✅ Assess each AI initiative against all 10 guardrails and apply controls proportionate to its risks
📝 Update policies and procurement to reflect supplier alignment with the guardrails
🔄 Integrate testing, documentation and oversight into your normal change-management
📅 Review systems at least annually or on material change

SME-scaled implementation approach¶

While the 10 guardrails apply to all organisations, SMEs can adopt them at different maturity levels. The levels below are a practical SafeAI-Aus scaling model, not part of the official VAISS.

Guardrail 1: Establish, implement and publish an accountability process

Minimum: Designate an AI responsible person
Better: Create simple AI governance policy
Best: Regular board/leadership AI updates

Guardrail 2: Establish and implement a risk management process

Minimum: Complete a documented AI risk and impact assessment
Better: Review risks on material change and at a defined cadence
Best: Integrate AI risks and controls into enterprise risk management

Guardrail 3: Protect AI systems and implement data governance measures

Minimum: Apply existing privacy, cybersecurity and data governance controls
Better: Add controls for AI-specific data, model and prompt risks
Best: Verify data provenance, access controls and security measures throughout the AI life cycle

Guardrail 4: Test AI models and systems

Minimum: Test against documented acceptance criteria before deployment
Better: Monitor performance and review after incidents or material changes
Best: Use ongoing, risk-proportionate evaluation and independent review where warranted

Guardrail 5: Enable human control or intervention

Minimum: Define when and how a person can pause, override or escalate the system
Better: Require meaningful human review for significant decisions
Best: Design and test risk-proportionate human oversight across the life cycle

Guardrail 6: Inform end-users regarding AI-enabled decisions

Minimum: Tell people when they interact with AI or receive AI-generated content
Better: Explain the AI system's role, capabilities and limitations in plain language
Best: Tailor accessible explanations and disclosures to affected stakeholder needs and risk

Guardrail 7: Establish processes for people to challenge use or outcomes

Minimum: Provide a channel to question or complain about AI use or outcomes
Better: Document response, escalation and correction processes
Best: Offer review independent of the original AI-assisted decision where the risk warrants it

Guardrail 8: Be transparent with other organisations across the AI supply chain

Minimum: Ask suppliers for system purpose, limitations, data and testing information
Better: Put information-sharing and incident obligations in contracts
Best: Use ongoing assurance or audit for material suppliers and higher-risk systems

Guardrail 9: Keep and maintain records

Minimum: Maintain an AI inventory with an owner, purpose, risk rating and review date
Better: Record assessments, controls, tests, incidents and approvals
Best: Maintain evidence proportionate to external assurance and regulatory needs

Guardrail 10: Engage your stakeholders and evaluate their needs

Minimum: Identify people and groups who may be affected
Better: Seek feedback during design, deployment and review
Best: Use inclusive engagement or co-design where impacts or stakeholder needs warrant it

Regulatory outlook and the future of the guardrails¶

As at 22 July 2026, Australia still has no general AI Act and VAISS remains voluntary.

The National AI Plan retains existing technology-neutral laws and sector regulators — including privacy, consumer law, financial services, safety and anti-discrimination frameworks — as the foundation for managing AI-related harms. The Australian AI Safety Institute is operating within the Department of Industry, Science and Resources to analyse AI capabilities, risks and harms and support regulators and agencies.

In July 2026, the Government separately announced plans for Australian Standards for AI. Published material focuses on a proposed framework for large data centres and AI training, including future mandatory requirements for large AI data centres. The final scope, duties and commencement arrangements remain subject to design, National Cabinet consideration and legislation.

The 2024 consultation on mandatory guardrails for high-risk AI remains useful policy history, but its proposed guardrails and definition of high-risk AI are not current legal requirements.

In practical terms, organisations should:

treat the 10 guardrails and AI6 as voluntary guidance, not proof of legal compliance
focus on integrating these controls into existing privacy, risk, safety, security and compliance processes
monitor the Office of AI, AI Safety Institute, NAIC and relevant sector regulators for changes.

Assess risk by use and context¶

The 2024 mandatory-guardrails proposal is not a current legal risk classification. For operational decisions, use the current AI6 screening and risk-assessment approach instead.

Do not assume a use is lower-risk because the technology is common. Assess the specific purpose, data, affected people, degree of human reliance and the likelihood, severity and scale of harm. Reassess when the system, use or operating context changes.