The Ombudsman has found serious failures in automated aged-care assessments. Labor promised to learn from Robodebt, yet human judgement remains sidelined when decisions affect older Australians’ care.
The Commonwealth Ombudsman’s investigation into automated aged-care assessments poses an uncomfortable question for the Albanese government: what was the point of the Robodebt Royal Commission if its lessons could be so readily disregarded?
Under the Support at Home arrangements introduced in November 2025, the Integrated Assessment Tool (IAT) uses an algorithm to determine funding classifications. Assessors collect information about older people’s circumstances but cannot override the resulting classification, even when their professional judgement indicates it is clinically inappropriate.
On 26 October, the ombudsman reported serious shortcomings in governance, transparency and safeguards, finding inadequate risk assessment and evaluation of the system deployed. The department disputes that characterisation. Six recommendations were accepted, but the recommendation to permit human overrides was only noted.
These findings expose more than a flawed computer program. They test whether Labor’s commitment to accountability extends to its own administration.
What Robodebt should have taught us
Established by Labor in 2022, the Robodebt Royal Commission examined an unlawful debt-recovery scheme administered under successive Coalition governments. Commissioner Catherine Holmes documented failures of legality, public administration and basic humanity.
Income averaging generated inaccurate debts. People receiving social security were expected to challenge government calculations they could not readily interrogate, bearing the emotional and financial costs of correcting official errors.
The commission’s recommendations extended beyond that particular scheme. Recommendation 17.1 proposed a consistent legal framework for automated government decisions, accessible review, understandable explanations and independent scrutiny of algorithms. Recommendation 17.2 proposed considering a body with powers to monitor and audit automated decision-making.
The aged-care system now displays familiar risks: opaque decisions, inadequate scrutiny and restrictions on human intervention.
The schemes are different. Robodebt pursued alleged welfare debts; aged-care assessments allocate support. Their legal questions and consequences differ. The comparison concerns governance: what happens when administrative systems cannot adequately explain their decisions or correct inappropriate outcomes?
Human assessment without human judgement
The government argues that standardisation can address inconsistencies in the previous Home Care Packages system and distribute resources more fairly. Unspent allocations have featured prominently in that defence.
There is a legitimate policy problem here. Public resources must be allocated fairly, and professional judgement is not infallible. An algorithm can help identify inconsistencies and support more reliable decisions.
As aged-care experts Kathy Eagar and Paul Sadler argued on P&I in July, unspent funds do not necessarily demonstrate that people received more money than they needed. Services may be unavailable, access delayed or money retained for changing circumstances. Those possibilities require investigation before unused allocations become evidence that people needed less support.
Nor does consistency establish accuracy. A system can apply the same inadequate assumptions to everyone.
If an algorithm helps determine access to essential care, its operation must be tested, its reasoning understandable and its outcomes meaningfully contestable. An assessor who recognises an inappropriate classification but cannot correct it has responsibility without sufficient authority. Human involvement in collecting information does not make the resulting decision accountable.
Why the problem persists
Hypocrisy alone is an insufficient explanation. Institutional pressures survive changes of government: rising expenditure, growing demand, workforce shortages and expectations of administrative efficiency. Automation promises predictable decisions and manageable budgets. The danger arises when those objectives displace the purpose of the service.
Aged-care assessment should establish what support someone needs to live safely and with dignity. Decisions about available resources should then make any shortfall explicit and identify how it will be addressed. When a funding classification becomes the effective boundary of care planning, the system risks allowing its allocation to define the person’s needs.
Robodebt demonstrated how flawed assumptions become dangerous when embedded in administrative machinery and insulated from challenge. Political commitments to fairness cannot compensate for weak safeguards in implementation. A royal commission changes little unless its findings alter everyday decisions, institutional incentives and lines of responsibility.
What accountability requires
The government has commissioned an independent review of aged-care assessment and prioritisation, due to report in the first half of 2027. That is welcome, but does not remove the need for immediate safeguards.
First, qualified assessors should be able to override algorithmic classifications where documented evidence justifies doing so. Reasons should be recorded and decisions audited. This would make professional discretion accountable while preserving scrutiny of inconsistent practice.
Second, the government should publish understandable explanations of the tool, independently validate the version in use and report on errors, review outcomes and differences between groups of older Australians. Transparency must help people understand and challenge decisions affecting their care.
Third, ministers and departmental leaders should explain why implementation proceeded without adequate safeguards, what advice informed those decisions and who is responsible for correcting deficiencies. Announcing another review cannot substitute for answering those questions.
Finally, the government should complete a coherent framework for automated decision-making across public services. Each new algorithm should face established requirements for legality, validation, transparency, review and human responsibility.
The test of Labor’s commitment
Labor was justified in condemning Robodebt. Its royal commission exposed how government can lose sight of people behind the data and allow damaging practices to persist.
The aged-care controversy shows why those lessons must apply across political parties and portfolios. Standardisation cannot substitute for judgement, and a review pathway offers limited protection if it cannot remedy the underlying error.
Older Australians should not have to master an opaque assessment system to obtain appropriate care. Nor should they have to wait for another royal commission to establish who was responsible.
Robodebt’s lasting lesson is that automated decisions affecting people’s lives must remain lawful, transparent, contestable and subject to meaningful human responsibility. If Labor applies that lesson only when judging its opponents, it has failed to learn it.
Komla Tsey
Komla Tsey is a Ghanaian–Australian writer, retired Professor of Education for Social Sustainability, and part-time Professorial Research Fellow at the Jawun Research Institute, Central Queensland University. His writing explores education, democratic culture, identity, Indigenous empowerment, colonial legacies and the moral certainties that shape public life. His forthcoming memoir, Botoku Child, traces a journey from a small village in Ghana to universities and communities across the world, examining how inherited beliefs, curiosity and lived experience shape both personal lives and democratic societies.