2026 NDIS REFORM BILL:
Has Labor Learned Nothing From Robodebt? — The NDIS Automation Risk
Labor says it learned from Robodebt. But now, the Australian government is writing new automated decision-making powers into the NDIS reform.
This should stop the country in it's tracks. Not only is AI in government becoming the catalyst for mass government redundancies. The 2026 NDIS Reform concerns every person with disability, every carer, every advocate, and every Australian who remembers what Robodebt did to vulnerable people between 2015 and 2019.
Because NDIS decisions are not simple calculations.
They are not clean data points.
They are decisions about disability, functional capacity, fluctuating support needs, safety, housing, trauma, communication barriers, carer burnout, family breakdown, behavioural risk, complex health needs, and survival.
And once those decisions are automated, the harm can move faster than people can appeal it.
The warning in the NDIS Bill
The Saturday Paper recently reported that the NDIS reform Bill would allow unprecedented automated decision-making, not just for simple administrative calculations, but for decisions involving discretion, evaluative judgement, and even a “state of mind being formed.”
That is not minor automation, a calculator doing maths in the background.
That is automation being allowed into decisions that require judgement.
And judgement is exactly where disability support gets complicated.
Former Human Rights Commissioner Ed Santow has also warned that the proposed power is incredibly broad and raises grave concerns about safeguards. His concern is simple: NDIS decisions require careful consideration of individual circumstances and how disability affects a person’s life. (LinkedIn)
For the Albanese government, that is the point.
Disability is not a spreadsheet
A computer program cannot see the full story.
A computer program cannot see the child melting down behind closed doors.
It cannot see the parent who has not slept properly in years.
It cannot see the person masking through an assessment.
It cannot see the difference between “can do once” and “can do safely, repeatedly and sustainably.”
It cannot see the informal support system quietly collapsing around the participant.
It cannot see the difference between independence and survival.
And it cannot understand the fear many people will carry into NDIS I-CAN assessment after years of being disbelieved, reassessed, rejected, isolated and forced to prove the same disability over and over again.
That is why NDIS automated decision-making is so dangerous.
Because the NDIS does not make decisions about numbers alone.
It makes decisions about people’s lives.
The Robodebt lesson was supposed to be clear
Robodebt was a national warning about what happens when government uses automated systems against vulnerable people without adequate lawfulness, human judgement, transparency, and NDIS safeguards.
The Robodebt Royal Commission delivered its final report in July 2023. The Prime Minister’s own statement said the Royal Commission found Robodebt was “crude and cruel,” neither fair nor legal, and a costly failure of public administration in both human and economic terms. (Prime Minister of Australia)
The Albanese Australian Government later said it had agreed, or agreed in principle, to all 56 recommendations of the Royal Commission as part of restoring faith, integrity and trust in government. (Ministers' Media Centre)
So the question now is obvious.
If just three years ago, Mr Albanese told Australia that "Robodebt must never happen again", why is his government now expanding NDIS automated decision-making powers inside the NDIS legislation reforms Bill?
NDIS I-CAN assessment without safeguards is not reform
The government will say this is about efficiency. But efficiency is not automatically good. Efficiency without NDIS safeguards just means people can be harmed faster. Bad decisions can be made faster. Wrong supports can be removed faster. Appeals can become harder. And participants can be left trying to explain complex human realities to systems built to process categories, not lives.
That is not progress. That is risk transfer.
From government to participants.
From policy designers to families.
It shifts the risk of government error onto the participants, families and carers. The very people least able to absorb another administrative failure.
The problem is not technology
This is not an argument against all technology. Good technology can help. It can reduce administration. It can organise evidence. It can improve communication. It can support consistency. It can help participants and families manage complexity.
But technology should support human decision-making.
It should not replace human judgement in decisions that require discretion, context, lived reality and functional understanding.
The issue is not whether the NDIA uses technology. Technology is here now, and should be used to improve efficiency of structured systems.
But, the Australian government should never allow automation to move into decisions where human judgement must remain essential to the heart and soul of the Scheme.
The Human Rights questions Parliament must answer before the Bill is passed
Before any automated NDIS decision-making power is passed, Parliament should be forced to answer some very basic questions:
Who is accountable when the automated decision is wrong?
How will participants know automation was used?
What evidence will the system rely on?
How will fluctuating disability be understood?
How will masking, communication barriers, family breakdown or carer burnout be recognised by a Robo-assessment system?
What NDIS safeguards will stop automated errors from becoming life-changing harm on an exponential scale?
How will people appeal a bad decision made by automation?
What checks and balances, what quality controls, will be in place to protect the human participants from machine learning gone wrong?
And why should Australians trust this, after Robodebt?
These are not technical questions.
They are human rights questions.
They are administrative justice questions.
They are disability safety questions.
The bottom line
Labor says it learned from Robodebt. But writing broad NDIS automated decision-making powers into the NDIS suggests the lesson has not fully landed.
NDIS decisions are not simple maths. They are complex, human, functional decisions that affect safety, dignity, independence, housing, health, participation and survival.
If the government truly learned from Robodebt, it should not be asking how much more it can automate.
It should be asking what must should be NEVER automate, and how can we protect the human rights of NDIS participants by writing good laws.
Because when the automated NDIS reassessment gets disability wrong, people do not just lose funding.
They lose stability.
They lose trust.
They lose support.
They may even lose their lives.
And sometimes, they lose the fragile existence they were just managing to hold together.
NDIS Reform Bill is bad legislation.
So confident. So wrong.
We can help
NDIS reassessments are becoming more evidence-driven. Ability Pathways can help you prepare for NDIS I-CAN assessment properly, build your case, and advocate for the supports you need.