The Australian Government’s AI technical standard

In July 2025, the Australian Government’s Digital Transformation Agency (DTA) released the Technical Standard for Government’s Use of Artificial Intelligence (TSGUAI). The standard provides guidelines on best practices for the design, deployment, and use of AI systems in government agencies. 

The standards are a key enabler for the Australian Government’s AI capability. They also mark a significant step in Australia’s AI maturity, creating a ripple effect across all sectors, both public and private. 

Defining AI transformation in the public sector

The ultimate purpose of the standard is to support agencies in effectively implementing AI systems and leveraging them to deliver high-quality services. 

To achieve this, the standard guides agencies through the end-to-end process of implementing an AI system. To ensure practical guidance, it breaks the lifecycle of an AI system into three phases: 

• Discover: Agencies conceptualise, design and prepare AI systems for deployment. 
• Operate: Agencies implement AI systems and establish processes for their ongoing oversight and performance tracking.  
• Retire: Agencies responsibly decommission redundant AI systems.  

Within these phases are a series of statements outlining the necessary actions that agencies must take as part of their AI transformation. Each statement has a set of criteria, marked as either: 

• Required: Legislation-driven criteria that agencies must meet to satisfy the statement. 
• Recommended: Criteria that are not a mandatory requirement to satisfy the statement, but a best practice that agencies should still seek to implement.  

DTA’s General Manager of Digital Strategy, Lucy Poole, says the standard is a key enabler of the DTA’s mission: to position Australia as a global leader in the safe and responsible adoption of AI, without stifling adoption.  

In doing so, it complements existing policies and frameworks governing AI in the public sector, including the Australian Government’s AI Ethics Principles, Policy for the Responsible Use of AI in Government, AI Assurance Framework, and the Voluntary AI Safety Standard. 

Building AI capability: An extension, Not an add-on

For public sector technology leaders, embedding AI into the fabric of government operations in a way that’s ethical, explainable, and accountable presents a new challenge. But it’s not all unfamiliar territory. 

The AI Technical Standard’s requirements largely reflect existing IT practices, many of which are stock standard. As the standard aims to be complementary to existing standards, there shouldn’t be any additional requirements for organisations that have already adopted the Australian Privacy Principles (APPs). As such, we typically ask the same questions of an AI solution that we would for a non-AI solution. 

Of course, there will be AI-specific considerations that are relatively new for organisations to grapple with. These include managing data, managing AI models, and achieving an organisational consensus on topics like ethics in AI. 

Instead, the key challenge for agencies will lie in the urgency to implement AI efficiently, effectively and responsibly. As AI evolves at unprecedented rates, and is further embedded in workflows across every function, government agencies can’t afford to delay in these areas.   

The impact of the AI technical standard on the private sector

While primarily designed for the public sector, the AI Technical Standard offers a valuable benchmark for AI maturity that private sector organisations may choose to adopt as a guide. Understanding the standard is especially pertinent for current and prospective public sector technology vendors. 

By now, all organisations understand the transformative impact of AI, but still struggle to measure its impact. As adoption accelerates, procuring organisations will become increasingly conscious of the additional costs of AI, and the benefits they expect in return. Eventually, agencies will expect a certain level of AI built into every solution. Organisations that can best demonstrate the tangible value (and ROI) of their AI solution will come out on top.  

We’ve been here before, most recently with cybersecurity. For a long time, cybersecurity tended to be an add-on. Now it’s table stakes, and every organisation requires a robust set of cybersecurity controls in every solution. Similarly, mandatory compliance has really taken hold. There are several mandated cybersecurity requirements that vendors must demonstrate compliance with to secure government work. One such example is the Essential 8, which has become a widely accepted set of cybersecurity controls that agencies require their vendors to comply with. Furthermore, agencies are increasingly mandating other controls, such as ISO 27001, and SOC audits. 

As government agencies adopt the AI Technical Standard, so too must any organisation that delivers AI to government agencies. The Digital Transformation Agency’s AI Model Clauses (v2), released in March 2025, offer a framework for how government agencies should procure AI services. These clauses also apply where a vendor uses AI to assist in delivering another service. Therefore, the AI Model Clauses should be considered gospel for any organisation looking to provide technology services to government agencies.   

AI: Another tool in the toolbox

Given its disruptive nature, it may be tempting to treat AI as a completely new frontier. However, the standard rightly emphasises embedding AI practices within existing governance workflows. Doing so will be crucial in ensuring agencies can efficiently establish, accelerate and future-proof the public sector’s AI capabilities. 

At the end of the day, AI is just another technology solution, albeit a revolutionary one. But it shouldn’t be the cause of organisational inefficiencies created by siloed governance forums. Agencies should instead integrate AI governance into an existing business process. This approach can accelerate AI’s acceptance and adoption, as it’s done with new technologies in the past.   

Of course, every organisation is different. Some organisational structures may necessitate specialised committees to oversee the mechanics of AI data, compliance, and ethics. In this case, their role should still be to provide recommendations to an overarching governance framework. 

The Importance of strategic partnerships in AI adoption

Implementing IT solutions is the easy part. The challenge lies in the preparation: considering enterprise architecture, ethics, policies, and the changes to human and business processes involved. 

Many agencies are unlikely to have the internal capability to deliver on all aspects of the standard. For example, running an ethics workshop to align senior executives requires highly specialised facilitation skills. Furthermore, not every agency has the necessary in-depth data capability to implement AI effectively and responsibly.  

These skills gaps present a significant risk to the strategic imperative for government agencies to adopt AI at pace. To solve this challenge, organisations will need a long-term AI partner who can deeply integrate into their operations, understand their unique needs, and facilitate smoother transitions and implementations over time. This specialised partnership will make the journey of AI adoption more manageable and effective, delivering on the standard’s promise. 

Looking for a partner to guide you through your agency’s AI journey? Interactive’s transformation specialists can help you develop a clear strategy and execute it with confidence. 

Contact us to find out more and get started with your AI transformation.