Evaluating the DTA's AI Model Clauses (v2.0 - March 2025): A Legal Insight for Australian Government Buyers

The March 2025 update to the Digital Transformation Agency's AI Model Clauses represents a timely and essential advancement in the Australian Government's approach to the Artificial Intelligence (AI) procurement. As AI becomes integral to delivering services, guiding decision-making, and streamlining internal operations, those clauses offer essential legal and ethical frameworks for responsible adoption.

At Synergy Law, we've assessed the updated clauses through a commercial legal lens. This article highlights the strengths of the revised clauses, areas where further enhancement, and explains how we can support government agencies in adopting, refining, and embedding these provisions in a way that manages risk without stifling innovation.

What Works Well

1. Strong Ethical and Practical Foundations
   The clauses now closely reflect the Australian Government's AI Ethics Principles - prioritising transparency, fairness, accountability, and privacy into procurement. Rather than relying on broad commitments, they mandate concrete proof of alignment from suppliers, encouraging proactive risk planning from the outset.
2. Lifecycle and Risk Focus
   Version 2.0 introduces robust obligations spanning the entire AI system lifecycle - from design and training to deployment and decommissioning. This comprehensive perspective prompts agencies and vendors to continuously track model performance, anticipate drift, and refine data and algorithms as circumstances evolve.
3. Enhances Transparency and Explainability
   For AI systems that influence rights or access to services, suppliers must guarantee that decisions are explainable to human reviewers. This is vital for ensuring accountability, fairness and compliance with administrative law.
4. Seamless Procurement Capability
   The modular design of the clauses allows for effortless integration into Commonwealth Contracting Suite and agency-specific procurement templates, giving legal teams the flexibility to embed them without compromising clarity or compliance.

Areas for Further Improvement

1. Tailored Risk Proportionality
   While designed to be universally applicable, the clauses don't yet provide sufficient distinction between low-risk and high-risk AI systems. Without further granularity, smaller-scale procurements may face disproportionately heavy compliance burdens.
2. Clarifying Audit and Oversight Provisions
   Clauses that require supplier cooperation in audits lack detail around triggers, enforcement mechanisms, and accountability. Agencies would benefit from more explicit guidance that informs resourcing and oversight planning.
3. Security and Data Sovereignty Gaps
   Though references are made to frameworks like the PSPF and ISM, the clauses do not go far enough in addressing offshore processing or cross-border data risks - critical issued in projects involving sensitive or citizen-centric data
4. Unclear Liability Structures
   Legal liability for AI caused harm - such as biased outcomes or system failures - is still underdeveloped. The clauses fall short in clearly defining fault attribution, compensation mechanisms, and dispute resolution protocols.
5. Integration with Other Legal Obligations
   Agencies will need to align these clauses with existing legal regimes - the Privacy Act 1988, FOI Act, and other emerging legislation. Without tailored legal advice, overlaps or interpretive conflicts may emerge.

How Synergy Law Can Assist

As part of Synergy Group, Synergy Law offers unique experience at the intersection of AI, law, and procurement for government clients. Our support includes:

• Customising model clauses to reflect the risk profile, impact and scale of your AI system
• Drafting and reviewing Statements of Work to ensure AI compliance is woven into key delivery milestones
• Supporting supplier engagement, from drafting clarifications to managing negotiations
• Providing ethics and privacy guidance to align with legislative frameworks
• Delivering training and internal briefings tailored for legal, procurement, and project teams

Whether your agency is deploying automation tools, chatbots, or advanced data analytics, we help craft contracts that are legally robust and practically tailored.

Final Word

The DTA's AI Model Clauses v2.0 represents a significant step forward - establishing a firmer foundation for safe, ethical AI in government procurement. However, they aren't a one-size-fits-all solution. To achieve AI implementations that are lawful, transparent, and effective, careful customisation is critical.

Synergy Law empowers agencies with legal clarity, hands-on experience in government contracting, and deep AI procurement insight. We're committed to helping you navigate this evolving landscape - so you get it right