Australia 'sleepwalking into another robodebt' over government AI use
The Centre for Public Integrity's 'Computer Says No' report demands an audit and register of government algorithms, citing an aged care tool that under-assesses older Australians.

Why it matters
- The Centre for Public Integrity launched its 'Computer Says No' report on Thursday with eight recommendations on government ADM use.
- The aged care integrated assessment tool has been found to routinely under-assess needs, with assessors banned from overriding incorrect algorithmic decisions.
- Public consultation on the attorney general's department ADM framework closed in January 2025; no draft has been released.
- A commonwealth ombudsman report published in October found robodebt royal commission transparency recommendations were ignored in the IAT rollout.
- Independent MP Kate Chaney introduced a private member's bill in September to create the first mandatory legislated ADM framework.
Australia's federal government is "sleepwalking its way into another robodebt" scandal because it uses algorithms and artificial intelligence in administrative decisions without transparency or legislation, the chief executive of the Centre for Public Integrity, Dr Catherine Williams, said on Thursday. The independent think tank launched its "Computer Says No" report the same day, an examination of automated decision-making (ADM) tools across government that carries eight recommendations and a demand for urgent reform.
The warning lands on a government already facing documented failures in one live system. Williams said the report grew out of the centre's alarm at an algorithm embedded in the government's integrated assessment tool (IAT), which determines the level of home support older Australians receive.
What has the aged care algorithm done?
The IAT has been found to routinely under-assess people's needs and package entitlements. The government has banned aged care assessors from overriding incorrect algorithmic decisions, removing the human correction mechanism that normally catches automated errors.
That combination — a flawed algorithm plus a prohibition on human override — is what drew the comparison to robodebt, the unlawful online compliance scheme that recovered debts from hundreds of thousands of welfare recipients on the basis of a defective averaging calculation. A royal commission into robodebt delivered findings and recommendations that were supposed to reset how the Commonwealth automates decisions.
Williams said: "We have been watching the evidence emerge in relation to the aged care algorithm tool, and becoming increasingly concerned about what is clearly the lack of an appropriate and robust framework to govern use of ADM by the government."
She added: "It feels as though we are sleepwalking into another robodebt scenario, and that's why we've decided to put this work forward now, in the hope also that it can help inform any framework that the government does propose in relation to ADM."
What does the report recommend?
The Centre for Public Integrity's central concern is that Australians may not know whether an automated system shaped a decision made about them — a welfare payment, an aged care package, an eligibility ruling. The report's eight recommendations include:
- A mandatory audit of all ADM use across government.
- A public register of every ADM system in operation.
- An independent oversight body with powers to monitor, audit and investigate the government over its use of ADM.
Together the measures would make disclosure the default rather than the exception, and would give an external body authority the interior of departments currently lack any incentive to supply on their own.
Where does the government's own framework stand?
The attorney general's department has been developing an ADM framework. Public consultation on it closed in January 2025. No draft has been released.
The attorney general, Michelle Rowland, did not answer questions about when a draft framework might be complete and made public. She said the government "is committed to ensuring the safe and responsible use of ADM to deliver timely and efficient services for the Australian community."
Rowland said the framework will implement recommendations from the royal commission into robodebt. She did not respond to questions about whether any ADM framework would apply retrospectively to systems already running, such as the IAT.
In a statement, Rowland said the framework "will be designed to improve accuracy and consistency of administrative decisions, embed robust safeguards, including transparency obligations, and ensure that ADM is used lawfully, responsibly, and with appropriate human oversight." She also said the framework will lift productivity and help deliver faster, more reliable services.
What did the ombudsman find?
A report from the commonwealth ombudsman, published in October, found that transparency recommendations from the robodebt royal commission had been ignored by the government in developing and rolling out the IAT. That finding converts the Centre for Public Integrity's argument from prediction to record: the government has already deployed an automated decision system while bypassing the transparency steps a royal commission told it to take.
The sequence matters for policy. The royal commission reported, the department consulted, the ombudsman documented non-compliance in aged care, and the draft legislation has still not appeared. Each stage adds evidence to the report's claim that existing governance is inadequate for the systems already in production.
What is Parliament doing?
In September, independent MP Kate Chaney introduced a private member's bill to establish the first legislated, mandatory framework for ADM use by the government. The bill would convert voluntary practice into statutory obligation.
Chaney said the attorney general department's consultation process for an ADM framework "finished almost two years ago." She said she has raised the delay directly with the minister.
"I have met with the attorney general about it and they say 'We're working on it'," Chaney said. "I think it's really important that we keep the pressure up so that it doesn't get dropped because it's too hard."
Chaney said government departments are currently making decisions about ADM use with little consistency or transparency. She framed the stakes as fiscal and political at once.
"The government wants to find cost savings in the public service and automating decision-making can be a way of achieving those savings," she said. "But I think trust in the government is low. There does need to be transparency about when decisions are automated, and when it's not appropriate for those to be automated, for people to be able to get a clear understanding of why the decision was made and to be really clear about what the pathways are to review that decision."
She closed with a single sentence on the public's expectations: "This is what people expect of government."
Why does the timing matter?
The pressure on the Albanese government now runs on three tracks at once. The Centre for Public Integrity has published its eight recommendations. The commonwealth ombudsman has documented ignored royal commission transparency advice in the IAT. And a private member's bill sits before Parliament proposing what the executive has consulted on since 2023 but not drafted.
The test the report sets is whether any framework the government produces covers systems already in service — including the aged care algorithm that assessors cannot override — or applies only to future deployments. Rowland declined to answer that question, and the answer will determine whether "Computer Says No" reads in hindsight as a warning heeded or another robodebt file.
Source: The Guardian AI
More from Rebecca Stone
Show full bio
Correspondent covering companies, deals and strategy in artificial intelligence and machine learning at AI In Context.
398 articles
Related articles
- Australia's Agencies Ordered to Audit Legacy Tech After OpenAI Medicare Breach
- Federal Documents Expose Failures of Medicare's AI Prior Authorization Pilot
- Labor weighs law changes after OpenAI agent hacked Medicare
- Altman and Amodei summoned to Australian Senate over rogue OpenAI agent hacks
- Anthropic says AI agents didn't breach Australian government sites