The warning that took two years to be heard
Somewhere inside KPMG Australia, someone raised their hand. According to Ethisphere, a whistleblower sent a detailed warning to leadership in May 2024 about senior partners misusing confidential client information. The firm handled it as a workplace grievance. Two reviews came back clean.
It took a senator using parliamentary privilege in March 2026 for the story to break open. By the end of May, the CEO had resigned. By August, almost 400 people were losing their jobs.
That gap between the warning and the reckoning is where the real cost of a poor speak-up culture lives. And it is far more common than most leadership teams want to believe.
The 2026 statistic every leadership team should read
In May 2026, Radical Candor published The Trust Gap: State of the Workplace Insights, based on a survey of 600 US workers conducted between February and March 2026.
The headline finding is uncomfortable. When asked whether they see people staying silent about problems, 48% of executives said yes. Employees put that number at 61%. Managers said 63%. HR teams said 67%.
Read that again. The people furthest from the problem see it least. The people closest to it see it most. And 45% of employees said they don’t feel safe at work at all.
This is the employee silence problem in plain numbers. Leaders are working from a picture of their organisation that is up to 19 points rosier than the one their HR teams see every day.
What happened at KPMG Australia
Here is what has been publicly reported, with sources.
In March 2026, Labor senator Deborah O’Neill shared a whistleblower’s allegations in parliament that confidential Lendlease board papers had been used to support KPMG’s bids for major audit tenders (ABC News, 24 August 2026).
On 29 May 2026, KPMG Australia confirmed that its treatment of the whistleblower and its investigation of their allegations had fallen short. CEO Andrew Yates and Julian McPherson, National Managing Partner for Audit and Assurance, resigned. KPMG said its initial investigations were not carried out with the rigour required (ABC News, 29 May 2026; KPMG statement, 29 May 2026).
On 16 June 2026, KPMG agreed to stop bidding for new federal government work until 30 September 2026. Its 297 existing federal contracts, worth A$653 million, were not included in the pause (ABC News, 16 June 2026).
On 24 August 2026, KPMG Australia announced it would cut 27 partners and about 360 staff, roughly 5% of its workforce. Revenue fell from A$2.28 billion to A$2.26 billion for the 2026 financial year. CEO John Sams said the cuts were a response to economic weakness, difficult market conditions and the impact of the firm’s conduct and whistleblower matters (ABC News, 24 August 2026).
KPMG’s response
KPMG has not disputed the core failure. Chairman Martin Sheppard said the firm would “apologise unreservedly to the whistleblower” and committed to reviewing its speak-up culture. In August, Sams said the firm recognised the challenges created by its own failings and that internal and external reviews were under way. It is fair to note that KPMG attributes part of its revenue pressure to wider market conditions, not the scandal alone.
Why this is a culture story, not just a compliance story
It is tempting to file this under audit ethics. But look at the sequence. Someone saw a problem. They spoke up. The system absorbed the complaint without acting on it. That is a psychological safety failure before it is anything else.
The Radical Candor data explains why this keeps happening. When 61% of employees see colleagues staying quiet about issues but leaders see far less, the organisation loses its early warning system. Problems don’t disappear. They just get more expensive.
The MoodyBot probability model: what KPMG Australia could have saved
MoodyBot, the AI layer inside Moody At Work, reads anonymous daily mood check-ins across teams and departments. It does not read documents, audit files or client data, and it could not have detected the data misuse itself. What it tracks is how people feel at work, every day, without their names attached.
In a situation like this one, the signals MoodyBot could have picked up include a sustained drop in mood scores inside a specific practice area, check-in reasons clustering around pressure, fear or unfairness, falling participation (people stop checking in when they stop believing anyone listens), and a widening gap between how a team feels and how its leadership feels.
With a warning reportedly sitting unresolved for close to two years, that is a long window for anonymous sentiment data to surface a pattern and put it in front of the board.
The model below uses only figures KPMG Australia or ABC News have made public. The attribution share and the early detection probability are illustrative assumptions, not facts, and are stated openly so anyone can challenge them.
| Risk | Publicly reported loss or exposure | Model assumptions | What MoodyBot could have saved |
|---|---|---|---|
| Revenue decline | A$20m fall in FY26 revenue (A$2.28bn to A$2.26bn) | 40% to 60% of decline linked to conduct matters; 40% to 60% chance early signals lead to action | A$3.2m to A$7.2m |
| Job losses | 27 partners and about 360 staff cut (about 5% of workforce) | Redundancy costs not publicly disclosed | Some roles could have been protected if the revenue hit was smaller; not quantified |
| Federal work | New federal bids paused 16 June to 30 September 2026; 297 existing contracts worth A$653m | Value of lost new bids not publicly disclosed | Earlier action could have reduced exposure; not quantified |
| Leadership exits | CEO and National Managing Partner Audit resigned | Replacement and disruption costs not publicly disclosed | Not quantified |
| Total quantified | A$3.2m to A$7.2m on the one line with a public dollar figure |
How the numbers were built: The only publicly reported dollar loss is the A$20 million revenue decline. KPMG itself named its conduct and whistleblower matters alongside market conditions, so the model assumes 40% to 60% of the decline is linked to the scandal. It then applies a 40% to 60% probability that anonymous early warning signals could have led to action in time. A$20m x 40% x 40% gives A$3.2m. A$20m x 60% x 60% gives A$7.2m. Every other cost remains undisclosed, so it is left out rather than guessed.
What every leadership team can take from this
You don’t need a Senate inquiry to find out whether your people feel safe speaking up. Ask yourself one question. If someone on your team saw something wrong today, how long would it take for you to hear about it?
If the honest answer is “I’m not sure,” the Radical Candor data suggests you are probably seeing less silence than your people are.
Frequently asked questions
Why did the KPMG Australia CEO resign? Andrew Yates resigned in May 2026 after KPMG confirmed its handling of a whistleblower’s complaint and the related investigation fell short of expectations.
How many jobs did KPMG Australia cut? In August 2026, KPMG Australia announced it would cut 27 partners and about 360 staff, around 5% of its workforce.
What percentage of employees see colleagues staying silent at work? According to Radical Candor’s 2026 Trust Gap report, 61% of US employees say they see people staying silent when there are issues, compared with 48% of executives.
How can companies detect a poor speak-up culture early? Anonymous, frequent check-ins give people a way to signal how they feel without risk. Tools like Moody At Work turn those signals into early warnings leadership can act on before problems escalate.
Sources: Radical Candor, The Trust Gap: State of the Workplace Insights (GlobeNewswire, 7 May 2026); ABC News (29 May 2026, 16 June 2026, 24 August 2026); KPMG Australia media statement (29 May 2026); Ethisphere (27 July 2026).
All factual claims are sourced from publicly available reporting. KPMG Australia’s representatives have stated that the firm’s treatment of the whistleblower and its investigation fell short of expectations, have apologised unreservedly to the whistleblower, and have commissioned internal and external reviews of the firm’s speak-up culture. MoodyBot projections represent an independent illustrative probability model based on publicly reported information and are not guaranteed outcomes.