The Audit That Found 1.6 Billion Euros of Unknown Benefit: Dutch AML Meets AI
The Netherlands Court of Audit reported in March 2026 that eight Dutch banks spent about 1.6 billion euros on roughly 13,000 anti-money-laundering staff in 2024 — and that the benefits are unknown while the harm to wrongly flagged citizens is documented. Banks are now automating the function rather than defending it.
Two documents published five months apart tell a story that neither one tells alone.
The first is from October 2025. At a finance conference in Driebergen, executives from ABN Amro, ING, Rabobank and ASN indicated they expect roughly 2,600 money-laundering-check positions to disappear within about two years, as reported by Het Financieele Dagblad. The second is from March 11, 2026, when the Netherlands Court of Audit published an investigation into what that money buys. Its finding: the eight banks it audited deployed roughly 13,000 full-time employees on anti-money-laundering work in 2024 at a cost of about 1.6 billion euros, up from 1.16 billion in 2021 — and the benefits of the regime are unknown, while the consequences for citizens wrongly caught by it are serious.
Put the two together and the sequence is not subtle. An occupation costing 1.6 billion euros a year, growing steadily, cannot demonstrate that it works. The people who run it expect a fifth of it to be gone inside two years.
The sourcing is worse than the headlines
Most coverage of this story has merged distinct announcements into a single "banks blame AI for AML layoffs" narrative. The record does not support that.
ING's March 2026 investor presentation projects a decrease of approximately 1,250 Operations full-time employees in 2026 — not a company-wide cut — with a footnote placing it "among others in KYC and associated activities, which account for approximately 6,000 FTEs." ING does not describe a 6,000-person AML division. It targets roughly 350 million euros of incremental cost efficiencies in 2026. Separately, ING filed with the Dutch benefits agency UWV in October 2025 that up to 950 Dutch positions may go by the end of 2026, attributed to digitalisation broadly, with AML unnamed. Those are two different numbers from two different announcements, and they are routinely printed as one.
ABN Amro's widely cited "35 percent of AML staff replaced by AI" is likewise a distortion. The bank's November 2025 announcement covers 5,200 cuts, about 20 percent of its workforce, by 2028; the roughly 35 percent figure spans customer service, operations and money-laundering checks together, and the AI attribution is an inference. ASN Bank is cutting up to 950 further roles and Triodos 250 to 270 by 2028, both general reorganisations rather than AML-attributed ones.
The accurate version is quieter and more interesting: compliance headcount has stopped growing for the first time since the fines that created it, and the reduction is arriving through attrition and unbackfilled requisitions rather than announced layoffs.
The negative business case
Jaap van der Molen, who leads financial crime work at ABN Amro, has called the regime a negative business case. The arithmetic is unkind: Dutch authorities seized on the order of 400 million euros in criminal assets in 2024 against roughly 1.4 to 1.6 billion euros a year of bank spend. Deterrence is real and hard to count, so that ratio is not the whole story — but the person running the function is saying publicly that the numbers do not work, and the national audit office is saying it cannot find evidence they do.
This is what makes AML different from every other function AI is currently eating. Automation usually has to prove it is as good as the human. Here, the output has never been measured, so the machine only has to be cheaper at producing the same documented diligence.
The regulator is not the shelter people think
The standard defense — the regulator requires a human — is weaker than assumed.
FinCEN's notice of proposed rulemaking on April 7, 2026 moves toward an effectiveness-based standard. It is explicitly not an AI mandate, but its fact sheet states that institutions responsibly experimenting with innovative technologies will not incur additional enforcement risk for doing so — removing the specific fear that adopting AI is itself an examination finding. FATF's horizon scan on AI and deepfakes stresses human oversight and explainability without prohibiting automated decisioning. The EBA handed its AML powers to the new EU authority AMLA on December 31, 2025.
And in the sharpest precedent, a Dutch court in litigation between the neobank bunq and the central bank DNB held that DNB was wrong to require manual transaction-monitoring processes. That ruling dates from 2022 rather than 2026, and its precise holding is worth checking before anyone leans on it — but the direction is the opposite of the assumption. A regulator was overruled for mandating humans.
The vendors have picked the category
On May 4, 2026, FIS announced a Financial Crimes AI Agent built with Anthropic, naming BMO and Amalgamated Bank as institutions developing with it, with general availability planned for the second half of 2026. The claims — investigations from days to minutes, fewer false positives, better SAR narratives — are unmeasured vendor marketing and should be discounted accordingly.
What is informative is the selection. When a core banking infrastructure vendor of that scale makes financial crime its opening agentic use case, the category has been chosen. One line in the announcement matters more than the speed claim: "Investigators will remain in control of every decision." That is a statement about accountability, not about who does the work — and it is entirely compatible with a tenth as many investigators.
What to watch
The pivotal moment will not be technical. It will be the first time a supervisor accepts a statistical sample of agent-authored dispositions as adequate review, the way model validation already works for credit decisions. That is the step that breaks the link between alert volume and headcount. Nobody has been examined on it yet.
The counterweight worth tracking: US Bureau of Labor Statistics projections show financial examiners growing about 19 percent through 2034, and compliance officers about 3 percent. Those broad categories include bank examiners and licensing staff who multiply when the rulebook grows. An occupational bucket can expand while its queue-clearing tier hollows out — which is exactly what the Dutch numbers describe.
For the full displacement analysis — the mechanism, the funnel, and what survives — see how AI will replace AML analysts, and the related prediction on where compliance headcount lands by 2027.