The Digital Transformation Agency (DTA) has launched a pilot programme to create an AI assurance framework aimed at ensuring ethical and responsible AI use across Australian government agencies. Starting in September 2024, this initiative is a crucial step in aligning government AI practices with the National AI Ethics Principles, helping to maintain transparency, fairness, and security in AI-driven public services. By prioritising human oversight and community welfare, the DTA addresses the risks inherent in AI deployment, strengthening accountability and resilience in AI-powered projects.
This pilot programme, a foundational element of the Australian Government’s AI policy, underscores a firm commitment to transparent and accountable public services powered by AI. As the framework develops, it aims to establish consistent standards for AI use within the government, advancing the DTA’s mission to lead in the safe, beneficial adoption of AI technology within Australia’s public sector.
AI assurance in government
The Digital Transformation Agency (DTA) has introduced a pioneering dual-assessment framework to guide responsible AI adoption across government agencies. This new process requires agencies to conduct an initial threshold assessment, where they closely examine the core objectives of any proposed AI use and evaluate whether non-AI alternatives may be more practical or cost-effective. “We want agencies to carefully consider viable alternatives,” said Lucy Poole, DTA’s General Manager of Strategy, Planning, and Performance. “For instance, non-AI services could be more cost-effective, secure, or dependable.”
If the threshold assessment reveals a moderate or higher risk level, agencies must then perform a comprehensive evaluation. This second assessment involves strict adherence to AI Ethics Principles, focusing on fairness, safety, privacy, and transparency. Agencies are required to address potential biases in training data to ensure outcomes are equitable and unbiased. They must also uphold stringent standards for data relevance and adhere to Indigenous data governance protocols, reinforcing AI’s commitment to safety and inclusivity. By establishing these strict assessment steps, the DTA ensures that AI technologies are implemented thoughtfully, safeguarding the interests of Australians and promoting a responsible AI landscape across government sectors.
Enhancing AI transparency standards
The Digital Transformation Agency (DTA) has reinforced its commitment to transparency and public accountability with a new AI assurance framework designed to guide Australian government agencies in using artificial intelligence ethically. This framework mandates a structured approach to stakeholder engagement, thorough documentation, and the disclosure of AI systems, ensuring all AI applications align with public interests and comply with regulatory standards. At the heart of this framework is inclusive stakeholder engagement, requiring agencies to actively involve diverse community voices in government AI initiatives.
By incorporating a wide range of perspectives, agencies can proactively address ethical considerations and respond to public concerns about AI. Additionally, the framework requires comprehensive documentation of AI systems, covering their design, decision-making processes, and performance metrics. This detailed record-keeping makes AI applications understandable and accessible to oversight bodies and the public alike.
The DTA’s framework further mandates that agencies openly disclose key information about their AI programmes, including transparency statements that outline AI’s purpose and its role in government decisions. This approach fosters public trust by making AI use visible and understandable to those it impacts. “Our goal is to provide a unified approach for government agencies to engage with AI confidently,” said Lucy Poole, DTA’s General Manager of Strategy, Planning, and Performance, underscoring the DTA’s dedication to ethical, accountable AI practices.
Strengthening AI governance practices
The Digital Transformation Agency (DTA) has introduced its AI assurance framework to elevate digital governance and data management within the Australian Government. This initiative, a cornerstone of Australia’s AI policy, positions the government as a frontrunner in ethical and responsible AI adoption. Through this framework, the DTA sets high standards for accountability, transparency, and public benefit across all government AI applications.
Rather than replacing existing regulatory measures, the framework enhances them, offering a cohesive approach for government agencies to effectively manage AI-related risks while fulfilling legislative mandates. The DTA’s pilot framework enforces structured assessments, addressing vital elements of data management, such as data privacy, security, and ethical usage.
Under this guidance, agencies must demonstrate adherence to data governance standards, rigorously safeguard data integrity, and promote responsible data use at every stage of AI deployment. The DTA mandates that government AI solutions comply with all “relevant legislative obligations,” including the Australian Privacy Principles and data minimisation standards. This approach aims to strengthen public trust in how government agencies handle data and apply AI technologies responsibly.
Enhancing AI framework adaptability
Following the completion of its initial pilot, the Digital Transformation Agency (DTA) will collect extensive feedback from participating agencies to enhance the AI assurance framework. The DTA will gather insights through targeted surveys, feedback sessions, and detailed interviews, ensuring diverse agency needs and operational contexts inform the refinement process. This collaborative feedback initiative will commence in November 2024, with ongoing engagement opportunities scheduled for late 2024 and early 2025.
This phased approach enables the DTA to make informed adjustments based on the real-world applications and challenges faced by government agencies. Lucy Poole, General Manager of Strategy, Planning, and Performance, stated that the framework is designed to evolve alongside the rapidly changing AI landscape.
She remarked, “Our guidance is iterative; it is meant to change and adapt based on the shifting AI landscape within the APS.” The DTA aims to ensure that the framework remains both flexible and robust, accommodating technological advancements and regulatory changes while upholding the ethical and accountability standards central to Australia’s AI policy.
The DTA will use the gathered evidence to shape its recommendations for AI assurance practices across the government, ensuring that Australia’s AI policies adapt in line with technological advancements.