The federal government has set out how it wants artificial intelligence to enter Australia’s aged care sector, and the roadmap is deliberately slow. Companion robots for lonely residents, apps that read pain from a wince or a change in behaviour, and systems that monitor activity around the clock all feature in the plan. So does a caveat: none of it will scale until small, tightly controlled pilots prove it is safe.
The signal came in the Department of Health and Aged Care’s contribution to the government’s April 2026 response to the Senate Select Committee on Adopting Artificial Intelligence, tabled in Parliament early that month. As Hello Leaders reported, the department outlined a suite of AI applications for a sector that now serves roughly 1.35 million Australians, while stressing that broad deployment remains some way off.
What the government is proposing
The proposed applications fall into three broad buckets. First, companionship robots and virtual agents aimed at easing loneliness and social isolation among older Australians. Second, continuous behaviour and activity monitoring designed to flag safety risks, such as a fall or a sudden change in routine, before they escalate. Third, pain-assessment and management apps that draw on behavioural cues and resident input to detect discomfort in people who cannot easily communicate it themselves.
Before any of that reaches a residential facility or a home-care client at scale, the department says it will run limited pilots involving up to 20 health professionals testing systems in controlled environments. The stated purpose is to review research, safety protocols and ethical guidelines, then refine the tools on real-world feedback rather than release them and hope.
The roadmap sits inside a larger policy scaffold. It follows the National AI Plan released in December 2025 and the Aged Care Data and Digital Strategy, both of which frame AI as a lever against workforce shortages, social isolation and rising demand from an ageing population. The government’s broader response to the Senate committee leans on the language of safeguards, transparency and accountability, positioning care as one of several sectors where AI could close service gaps if handled carefully.
The shadow of the Integrated Assessment Tool
The caution is not abstract. It is shaped by a live controversy. In November 2025 the department launched an AI-assisted Integrated Assessment Tool, an algorithm-driven system that standardises eligibility and funding decisions for home-based care. It has not gone smoothly.
According to Hello Leaders, the tool has drawn more than 800 review requests since launch, alongside a reported 50 per cent increase in calls to advocacy services from people struggling to navigate their assessments. A central complaint is that the system’s decisions cannot be readily overridden by clinical judgement, prompting critics to reach for the most loaded comparison in recent Australian policy memory: Robodebt 2.0.
The Commonwealth Ombudsman is now investigating complaints about the tool, examining its transparency, fairness and usability. That probe running in parallel with the rollout plan is the clearest evidence of the tension at the heart of this story. The same department promising a measured, pilot-first future for companion robots and pain apps is already fielding an inquiry into an automated system it deployed less than a year ago.
A workforce crisis the technology cannot fix alone
The backdrop is a sector under structural strain. Australia faces a shortfall of at least 110,000 direct aged care workers within the next decade unless the workforce is boosted, the Committee for Economic Development of Australia has warned. Demand is rising faster than staffing, and care is becoming more complex. The Aged Care Act 2024, which commenced on 1 November 2025 to embed the rights-based reforms of the Royal Commission into Aged Care Quality and Safety, has raised expectations further, layered on top of the mandated 215 care minutes per resident per day.
That gap is exactly what the technology industry has moved to fill. The global “agetech” market is projected to reach A$170 billion by 2030, and its pitch is seductive: AI that eases loneliness, monitors safety and stretches thin rosters. But not everyone is convinced the framing is honest.
Writing in The Conversation, University of Sydney researchers Barbara Barbosa Neves, Alexandra Sanders and Geoffrey Mead argued that AI companies are selling a “false narrative” about fixing aged care. After analysing the marketing of 33 agetech firms across Australia, East Asia, Europe and North America, they found the industry tends to depict residents as trackable metrics and casts AI as the inevitable answer to what is really a social and political crisis. Their warning is blunt: AI does not eliminate work so much as shift it, and the relational parts of care that build trust get marginalised because they are hard to measure.
Why it matters for Australia
Aged care is one of the country’s highest-stakes settings for AI, precisely because the people it touches are among the most vulnerable and the least able to contest a bad automated decision. The upside is real: a pain app that catches distress in a resident with advanced dementia, or a companion agent that fills empty evenings, could genuinely improve lives. So could tools that give overstretched staff earlier warning of a fall.
But the downside is equally concrete. The Integrated Assessment Tool shows what happens when an algorithm mediates access to care and human override is unclear: hundreds of disputes, a spike in distress calls, and an Ombudsman at the door. For a sector still rebuilding public confidence after the Royal Commission, a second automation misstep would be costly. The government’s insistence on 20-professional pilots reads as a direct lesson learned, an attempt to test in a lab before repeating a mistake at national scale.
There is also a dignity question that no pilot can fully resolve. Substituting a robot for human company may ease a loneliness metric while quietly narrowing what care means, a concern the University of Sydney researchers press hard. The policy challenge is to capture AI’s efficiency without letting efficiency become the only thing that counts.
For now, the plan is a promise rather than a deployment. The measure of it will be whether the pilots are genuinely allowed to fail before anything scales, whether clinical judgement can override an algorithm when it matters, and what the Ombudsman concludes about the tool already in the field. If Canberra gets the sequence right, aged care could become a template for responsible AI adoption in a high-risk sector. If it does not, the Robodebt comparison will only get louder.
Sources: Hello Leaders, Department of Industry, Science and Resources (Government response to the Senate Select Committee on Adopting AI), The Conversation, Committee for Economic Development of Australia.
