For two years, a single Senate estimates exchange has anchored Australia’s NDIS fraud debate. In mid-2024, NDIA integrity chief John Dardo told a committee that 90 per cent of a certain group of plan managers showed “significant indicators of fraud” in their claiming behaviour.
The line has since been repeated by government ministers, opposition frontbenchers, journalists and, more recently, by campaigners pushing for the government’s amendment bill to pass the Senate. It has become one of the most durable facts in Australian political debate: nine in ten plan managers, fraudulent.
It is also, according to the industry body representing those plan managers, not quite what was said.
Disability Intermediaries Australia has pointed out that Dardo’s actual testimony was narrower than the version now in wide circulation. He was describing a subset of smaller operators, those managing funding for fewer than 100 participants each, out of roughly 1,500 active plan managers nationally.
The analysis behind the figure was based on claiming-pattern behaviour, not confirmed fraud. And the industry body notes that the overwhelming majority of participants who use a plan manager are served by a much smaller group of larger, established providers, not the long tail of operators the statistic was actually about.
None of that makes the underlying concern trivial. But it does mean one of the central numbers driving Australia’s current NDIS reform debate has been stripped of its context so many times that the version everyone is arguing over barely resembles the version that was said under oath.
This matters beyond pedantry. The NDIS Amendment Bill 2026, now stalled in the Senate pending a committee report due August 14, has become a flashpoint precisely because both sides are using fraud figures as ammunition. The government points to an estimated $11.5 million a day in fraud and integrity leakage lost to delay.
The Coalition accuses Labor of letting $5 billion a year in taxpayer funds go to criminals who have infiltrated the system. Advocacy groups warn that the bill’s new powers, including automated debt notices with no clear right of appeal, risk repeating the mistakes of Robodebt.
Almost none of this argument turns on whether the 90 per cent figure is accurate. It has simply become background noise, cited so often that its accuracy is now assumed rather than checked. That is a problem for public trust regardless of which side of the debate someone sits on.
If the scheme’s most repeated fraud statistic turns out to be a simplification of a much narrower internal finding, then the entire framing of “plan managers as the weak link” deserves a second look, not because fraud isn’t real, but because the picture of where it is concentrated may be wrong.
Here is what has had almost no scrutiny at all: the design feature that makes plan managers matter in the first place.
Only a small fraction of NDIS providers nationally, well under 10 per cent, are formally registered with the NDIS Quality and Safeguards Commission and subject to its audits. The rest, more than a quarter of a million providers, operate unregistered.
Unregistered providers can only be paid at all if a participant either self-manages their own funding or uses a plan manager to handle claims on their behalf. In other words, plan managers are not a peripheral part of the scheme’s integrity architecture. They are the financial gateway that allows the vast majority of an almost entirely unaudited provider market to draw down public money.
That is a genuinely significant structural fact, and it has barely featured in coverage of the reform bill at all. The current legislative debate is overwhelmingly focused on participant-facing measures: tightening eligibility, expanding debt recovery, introducing automated decision-making to catch irregular claims faster.
What it does not appear to do, at least based on public summaries of the bill, is meaningfully strengthen registration or audit requirements for plan managers themselves, the layer that determines whether hundreds of thousands of otherwise unaudited providers get paid.
If the government’s own framing is that plan managers are where a large share of “significant fraud indicators” cluster, it is a fair question why the reform aimed at fixing the scheme’s integrity problems does relatively little to change how that particular gateway is audited or regulated.
The bill tightens who gets into the scheme and how debts get recovered from participants. It says comparatively little about who gets to control the money once someone is in.
A proper version of this story would need a few things this piece cannot supply on its own. It would need Disability Intermediaries Australia on the record again, with a current-dated response addressing the 2026 bill specifically rather than their 2024 statement.
It would need updated NDIA figures on plan manager registration numbers, since the 1,540 figure still circulating is from an earlier estimates round. And it would need the primary Hansard transcript of the original exchange between Senator Steele-John and Dardo, rather than relying on the secondary paraphrases that have done the rounds since, given how contested even that paraphrase now is.
What is already clear is this: Australia’s NDIS fraud debate has spent two years fighting over a number that may not mean what almost everyone citing it thinks it means, while the actual chokepoint that number was trying to describe, a lightly audited layer sitting between billions of dollars and an almost entirely unregistered provider market, has had almost no scrutiny of its own design.
