THE $31.2M BIOMETRIC BONFIRE: WHO KEPT CABINET IN THE DARK?
The failed biometrics programme is being written off at a cost of $31.2 million, but the bigger question is whether officials kept decision-makers in the dark.
A failed Immigration New Zealand biometrics project has landed taxpayers with a $31.2 million Budget 2026 write-off and triggered an integrity investigation after Immigration Minister Erica Stanford alleged officials deliberately kept ministers in the dark.
The allegation is serious. It is also, at this point, an allegation. But it shifts the story beyond the familiar sigh of another government technology project going sideways. Stanford told a parliamentary committee during Scrutiny Week that officials “deliberately withheld” information, used “creative accounting” and split work to avoid Cabinet scrutiny, according to RNZ reporting. MBIE chief executive Nic Blakeley has apologised and accepted accountability on behalf of the ministry, while Public Service Commissioner Sir Brian Roche is investigating integrity concerns raised by an independent review.
That makes the central question simple: did the Biometric Capability Update fail because people made bad calls, or because the oversight system was dodged, misled, or too weak to notice?
What Was Being Built
The formal project was the Biometric Capability Update, or BCU, run by Immigration New Zealand. It was intended to replace ageing biometric identity systems and improve how fingerprints, facial images and other biometric data were collected, stored and used.
Immigration NZ said the upgrade would “future-proof” identity management and help manage immigration risk. In practical terms, it was meant to speed up identity verification and fraud detection. RNZ reported last year that one intended benefit was automated photo matching, reducing the roughly 15 percent of checks staff were doing manually.
It did not deliver. MBIE now says the project was discontinued in November 2025 after significant challenges made it no longer viable, so the old systems remain in use.
Seven Years, No Delivered System
The public timeline begins in November 2018, when MBIE says the BCU project was launched. According to later reporting and review material, the initial business case put the whole-of-life cost at about $19.5m.
In 2020, Immigration NZ shifted toward an off-the-shelf commercial solution. That may sound like the safe, sensible option — the public-sector version of buying from the shelf rather than asking someone to invent a shelf. But the project’s costs and risks kept rising.
By mid-2023, the forecast whole-of-life cost was about $35m, including support and maintenance. In July 2023, RNZ reported, two joint ministers approved spending up to that amount. Immigration NZ later said Cabinet approval was not required because Cabinet sign-off was triggered only if whole-of-life cost exceeded $35m.
That $35m figure is now the hinge of the whole governance story.
After the 2023 election, Stanford became Immigration Minister. In 2024 the project was still alive, but sick. RNZ reported in October 2024 that go-live had been pushed from late 2024 to April 2025 and that the delay would cost an extra $1.3m, taking the total to $36.3m. A March 2024 ministerial briefing reported by Biometric Update said costs had risen from $19.5m to $35m and warned they may need to rise to $40m.
By November 2025, MBIE had pulled the plug. Budget 2026 then recorded the damage: Stanford said the Budget included a $31.2m write-off for the project. The research brief identifies this as an impairment of intangible assets in Vote Immigration, Citizenship and Border Management — accountant-speak for money spent on an asset now judged to have no recoverable value. The public gets a line item. MBIE gets a review. The system gets nothing.
The Cabinet Threshold Problem
The most troubling issue is not simply that the project came close to the $35m Cabinet threshold. It is what officials did, or did not do, as the project approached and apparently crossed it.
RNZ previously reported Immigration NZ’s position that the BCU had an approved whole-of-life cost of up to $35m and therefore did not need Cabinet approval. Stanford now alleges the project was split into two parts and that “creative accounting practices” were used to keep costs below Cabinet’s mandated limit.
No public document yet proves an intentional strategy to avoid Cabinet. A project can be phased for ordinary delivery reasons. Costs can be categorised badly without someone twirling a moustache in a spreadsheet. But the independent review’s reported findings sharpen the concern: ministerial reporting was described as inconsistent and, at times, misrepresenting the true status of the project. Stanford also said advice she later received turned out to be “complete fiction”.
If accurate, that is a breakdown in the basic bargain of public administration: officials advise, ministers decide, Parliament scrutinises, and the public pays only after someone with authority has actually been told what is going on.
The Cabinet Manual expects public servants to keep ministers fully informed of significant developments and provide free and frank advice. A seven-year technology project rising from roughly $19.5m to $35m, possibly $40m, while failing to deliver measurable benefits would seem to qualify as significant by any definition short of “asteroid impact”.
So Who Knew What, And When?
Publicly available documents remain thin. MBIE has released its statement and the independent review. RNZ and specialist reporting have referred to ministerial briefings and internal warnings, including an independent review warning a year earlier that the original go-live was unlikely. But key documents: business cases, risk registers, project board minutes, gateway reviews, procurement papers and Cabinet advice, (if any) are not publicly available.
The responsible immigration ministers over the project’s life included Iain Lees-Galloway, Kris Faafoi, Michael Wood, Kiri Allan and, from late 2023, Erica Stanford. On the public record, the clearest formal approval was the July 2023 delegated approval up to $35m. There is no public Cabinet paper showing Cabinet considered the project as costs escalated.
Stanford says she did not receive the independent review until mid-June 2026, even though the project had been stopped months earlier. MBIE has not publicly admitted deliberate deception. Blakeley’s position is institutional: the project failed, that failure was unacceptable, MBIE fell short, and employment matters may be considered after the Public Service Commission process.
It’s worth remembering - an apology is not the same as a confession.
Where The Money Went
The known figures overlap but do not yet reconcile neatly.
MBIE has referred to a $33m project that failed to deliver. Budget 2026 includes a $31.2m impairment. Earlier figures put the original business case around $19.5m, the approved whole-of-life cost at $35m, and possible future cost at $40m. RNZ has also reported the possibility of further missing costs, including references to another $8m, though the public record does not yet clearly show whether that is part of BCU, related identity work, or a separate cost.
The main vendor named publicly was NEC New Zealand, with Datacom providing system support and Argonaut associated with the Secure Real Time Platform component. Contract values, variations, termination terms, internal labour costs and any reusable assets have not been fully disclosed.
What Roche Is Investigating
Public Service Commissioner Sir Brian Roche has said the integrity matters highlighted by the review are serious and concerning, going to the core behaviours and ethics expected of public servants. He plans to appoint an independent investigator. At Stanford’s request, he will also review the broader Our Future Services immigration programme to test whether ministerial advice can be relied upon.
No final findings are available. No staff member has been publicly disciplined in connection with BCU. MBIE says it will consider employment matters after the commission’s work.
Decorative Oversight, Or Dodged Oversight?
The evidence already supports a finding of bad delivery and weak governance. MBIE says as much. The unresolved question is whether the failure also involved deliberate avoidance of scrutiny.
That cannot be settled by ministerial anger alone, even when the anger appears well founded. It requires documents: what internal project boards knew, what assurance reviews warned, what ministers were told, how costs were categorised, and whether anyone discussed keeping the project under Cabinet thresholds.
For taxpayers, the immediate result is blunt: more than $30m spent, a $31.2m write-off, no new biometric system, and ageing infrastructure still needing maintenance. For democratic scrutiny, the result may be worse: If a major technology project can drift for seven years, brush a Cabinet threshold, fail to deliver, and only become visible when the money is already gone, then oversight is not operating as advertised.
Whether it was decorative oversight or dodged oversight is now for Roche to test. Either way, the bonfire has already been lit.