The short version
- The Centre for Public Integrity argues that current government practices regarding artificial intelligence lack necessary transparency and oversight mechanisms.
- Evidence suggests the integrated assessment tool for aged care routinely underestimates support needs while preventing human assessors from correcting algorithmic errors.
- Legislative efforts to mandate independent audits and public registers of automated systems face delays, with officials citing ongoing framework development.
Federal authorities in Australia are facing renewed scrutiny over their reliance on automated decision-making tools, with critics warning that the administration is drifting toward a recurrence of the widespread errors seen during the previous robodebt scandal. Dr. Catherine Williams, chief executive of the independent thinktank Centre for Public Integrity, characterized the current trajectory as a dangerous lack of legislative guardrails surrounding the use of algorithms and artificial intelligence in public sector operations. Her organization released a comprehensive report titled Computer Says No on Thursday, detailing urgent concerns about how these technologies are deployed without adequate transparency or accountability structures in place.
The core argument presented by the Centre for Public Integrity is that citizens remain largely unaware when automated systems influence decisions that directly impact their lives. The report outlines eight specific recommendations aimed at rectifying this opacity, including the establishment of a mandatory audit process and a public register documenting all instances where government agencies utilize automated decision-making tools. Furthermore, the thinktank advocates for the creation of an independent oversight body empowered to monitor, audit, and investigate governmental use of these technologies, ensuring that administrative actions remain subject to rigorous external review.
Immediate concerns have been triggered by the performance of the integrated assessment tool, a system used to determine the level of home support entitlements for older Australians. Evidence indicates that this algorithm frequently underestimates the care needs of individuals, resulting in reduced package entitlements. Compounding the issue is a policy restriction that prohibits aged care assessors from overriding decisions made by the algorithm, even when those determinations appear incorrect. This inability to intervene has raised alarms among advocates who view the tool as a primary example of how automated systems can cause tangible harm when human oversight is effectively removed from the equation.
Williams emphasized that the decision to publish the report was driven by mounting evidence regarding the aged care algorithm’s shortcomings. She expressed deep concern about the absence of a robust framework to govern such tools, noting that the situation feels reminiscent of the conditions that led to the robodebt crisis. The timing of the release is intended to inform any future legislative proposals the government might introduce, aiming to prevent similar systemic failures from taking root in other areas of public service delivery.
In response to these criticisms, the attorney general’s department has indicated it is developing a framework for automated decision-making, though the timeline for its completion remains unclear. Public consultation on this matter concluded in January 2025, yet Attorney General Michelle Rowland did not provide specific dates for when a draft framework would be finalized or made public. She affirmed the government’s commitment to ensuring that these tools are used safely and responsibly to deliver efficient services, while also noting that the new framework will incorporate recommendations from the royal commission into robodebt.
However, questions remain regarding the scope and retroactivity of the proposed measures. Rowland declined to specify whether the upcoming framework would apply retrospectively to existing systems like the integrated assessment tool. She stated that the guidelines are designed to improve accuracy and consistency in administrative decisions, embed robust safeguards including transparency obligations, and ensure lawful use with appropriate human oversight. The government also argues that these frameworks will enhance productivity and help deliver faster, more reliable services to the community.
Critics point out that progress on regulatory standards has been slow. A report from the Commonwealth Ombudsman published in October found that the government had ignored transparency recommendations from the robodebt royal commission when developing and rolling out the aged care assessment tool. This finding underscores the gap between official commitments to reform and actual implementation practices, suggesting that without legislative force, voluntary compliance may be insufficient to protect citizens from algorithmic error.
Independent MP Kate Chaney has introduced a private member’s bill to establish the first legislated, mandatory framework for automated decision-making use by the government. She highlighted that the attorney general department’s consultation process ended nearly two years ago, expressing frustration with the lack of concrete progress. Chaney noted that while she has met with officials who claim work is ongoing, she believes sustained pressure is necessary to ensure the issue does not fade from the political agenda due to its complexity.
Chaney argued that government departments are currently making decisions about automated systems with little consistency or transparency. While automation may offer a pathway to cost savings within the public service, she warned that trust in government institutions is already low. She stressed the importance of clarity regarding when decisions are automated, why specific tools were chosen, and what pathways exist for individuals to review and challenge those decisions. Without such transparency, citizens remain unable to understand the basis of administrative actions that significantly affect their welfare.
The debate over automated decision-making in Australia reflects a broader tension between efficiency gains and the need for accountability in public administration. As algorithms become more embedded in service delivery, the risk of systemic bias and error increases, particularly when human oversight is constrained. The calls for mandatory audits and independent oversight bodies represent an attempt to balance technological innovation with democratic principles, ensuring that automated systems serve the public interest rather than obscure it.
Sources behind this briefing
Go to the original reporting
- The Guardian World↗Albanese government warned it’s ‘sleepwalking into another robodebt scenario’ over use of algorithms and AI
