Australia Warned Over AI and Algorithm Use in Government Decisions

The Australian government is being warned that it is drifting toward another robodebt-style scandal because of weak rules on the use of algorithms and artificial intelligence in official decisions. According to The Guardian, Dr Catherine Williams, who leads the independent thinktank the Centre for Public Integrity, made the claim when her organisation released its “Computer Says No” report on Thursday.

The report looks at how the government uses automated decision-making, often shortened to ADM, and calls for urgent reform. The centre says Australians may have no way of knowing whether an automated system played a part in a decision that affects them.

What the report recommends

The report contains eight recommendations. The most significant include a mandatory audit of all ADM use, a public register listing where such tools are used, and an independent oversight body. That body would have the power to monitor, audit and investigate the government’s use of these systems.

Williams said the work was prompted by growing concern about an algorithm used within the government’s integrated assessment tool, known as the IAT. The tool is used to decide how much home support older Australians are entitled to receive.

Concerns about the aged care tool

The IAT has reportedly been found to regularly under-assess people’s needs and the value of their care packages. Aged care assessors have been prohibited by the government from overriding algorithmic results they believe are wrong.

Williams said the centre had been following the evidence about the tool and had become increasingly worried about the lack of a strong framework to govern how the government uses automated decisions. She warned that the country could be heading toward another robodebt episode, and said the centre hoped its work would help shape any framework the government proposes.

The government’s response

Work on an ADM framework sits with the attorney general’s department, where a public consultation wrapped up in January 2025. Attorney General Michelle Rowland did not say when a draft framework might be completed and made public.

Rowland said in a statement that the government intends to use automated decisions in a safe and responsible manner. She said the framework would put into effect recommendations from the royal commission into robodebt. Earlier, in October, the commonwealth ombudsman released a review concluding that the government disregarded transparency advice from that royal commission while building and launching the IAT.

Rowland did not answer whether the framework would apply retrospectively. She said it would aim to improve accuracy and consistency, include transparency obligations and require appropriate human oversight. She also said it would help raise productivity and deliver faster services.

Parliamentary pressure

Kate Chaney, an independent MP, put forward a private member’s bill back in September that would set up the first mandatory, legislated structure for how government uses ADM. Chaney said the department’s consultation had ended almost two years ago. She said she had spoken with the attorney general, who told her the work was ongoing, and she urged that pressure be maintained so the issue is not shelved.

Chaney also said departments are currently making automated decisions with little consistency or openness. She acknowledged that the government wants savings in the public service and that automation could help deliver them. However, she said public trust in government is low, and that people need clear explanations of why a decision was made and clear routes to challenge it.

Frequently asked questions

Sources

• The Guardian: Albanese government warned it’s ‘sleepwalking into another robodebt scenario’ over use of algorithms and AI

This article was written with the help of AI from reports by the outlets listed above. See our Editorial Policy. Spotted a mistake? Tell us.