An algorithm is deciding older Australians' aged-care funding, and getting it wrong for one in five who challenge it

Almost one in five older Australians who challenged their aged-care funding decision had it changed in their favour once a human looked at it, according to government data. The same data shows the government has refused to hand over more recent figures.

Of 757 older people who asked for a review of their Support at Home funding as of 31 March, 141 got a better outcome. That means they became eligible for more services, a higher level of support, or a higher priority. Another 177 saw no change, 304 were closed, and 135 were still being processed. (The Guardian, 19 August 2026)

Support at Home is a program that helps older Australians get care in their own homes rather than moving into an aged-care facility. The Department of Health and Ageing would not provide more recent data on how many people have contested their funding under the system that decides it, called the integrated assessment tool, or IAT, which started in November. (The Guardian, 19 August 2026)

Here is how the IAT works. An assessor sits down with an older person and asks questions about their health, daily life, and social situation. That information goes into a computer algorithm, which decides how much support the person gets. The catch: the assessor is not allowed to change the computer's decision, even if they think it got things badly wrong. Imagine a doctor examining you but being forbidden to change a computer's diagnosis. (The Guardian, 19 August 2026)

The Commonwealth Ombudsman, an independent watchdog that investigates government decisions, started looking into the IAT in April. Assessors and older people have called the tool "cruel" and "inhumane." The investigation is still going. (The Guardian, 19 August 2026)

Calls for help with aged care jumped by 50% after the IAT was introduced. Critics have called it a terrible system that ignores professional judgment. (The Guardian, 23 February 2026) The Guardian reported in April that there is no legal basis for stopping assessors from overriding the algorithm, which often gives people less support than they need. (The Guardian, 3 April 2026)

Greens senator Penny Allman-Payne, who wants the government to let assessors override the computer, said the real problem is bigger than the numbers show. Many older people do not even know they can challenge a decision, she said. The IAT is under-assessing older people "far in excess" of what the review figures suggest. She also said the system cannot accurately assess people with cognitive issues, mental health conditions, or frailty, because those needs are hard to capture in a standard questionnaire. (The Guardian, 19 August 2026)

Mark Aitken, a nurse who has worked in aged care, went public in February about the lack of human oversight in home support decisions. He has since started a service called Mascotte Aged Care Solutions to help older people and their families navigate the system. Aitken said he has been helping people challenge their allocations, but warned that many older people do not have the luxury of time. (The Guardian, 19 August 2026)

The system has left many people without enough care, which could push them into aged-care homes sooner than necessary. (The Guardian, 19 August 2026)

A bill to let assessors override the computer passed the Senate in July. (Australian Ageing Agenda, 2 July 2026) The Department of Health says the tool is designed to be "robust and concise" for identifying older people's needs in residential aged care. (Department of Health, 8 July 2026)

The broader context here matters. The 141 successful challenges out of 757 are not trivial. That is a near-one-in-five overturn rate on a system the government says is working as designed. If Centrelink's algorithm produced that ratio on appeal, it would trigger immediate questions about whether the original decisions were sound. Aged care has not yet generated that level of political heat, though the Ombudsman's investigation and the Senate bill suggest the pressure is building.

Allman-Payne's point about under-reporting gets to the heart of the problem. The 757 review requests only count people who knew enough, or had enough help, to challenge. The IAT's design means an assessor can watch the computer get it wrong right in front of them and be powerless to fix it. That is not a glitch. It is a deliberate design choice, and one with no legal basis, according to The Guardian's April reporting. The people most likely to be misassessed, those with cognitive issues, mental health conditions, and frailty, are also the people least able to navigate a review process.

The department's refusal to provide data more recent than 31 March raises its own questions. Five months of review data are missing from public view. That is a long stretch in a program that only started in November. If the overturn rate has stayed the same or got worse, the government has a problem. If it has improved, there is no obvious reason to keep the figures quiet.