Restructuring at Dutch banks and AI in anti-money laundering and customer service

In February 2026 the Dutch employee insurance agency (UWV) took stock of the restructurings at Dutch banks. ABN AMRO is cutting a net 5,200 FTEs by 2028, ING and ASN Bank are each cutting around 900 jobs, and Triodos 250 to 270; the Dutch central bank (DNB) wants to save 10 per cent on costs through a reorganisation of its own. Cost saving is the common denominator, with a merger or the threat of a takeover as the trigger at some of the banks. Most banks cite AI as a way to do the same work with fewer people. In 2025 banks gave notice of intended collective redundancies for more than 3,000 employees, against a few hundred in 2024.

According to the banks, it mainly concerns work governed by fixed protocols: transaction and anti-money laundering checks, customer service queries and administration. Every bank has built up large anti-money laundering departments, partly in response to heavy fines in the past, and that work generates no revenue. The banks expect AI to make 20 per cent of jobs in this work redundant within two years; ABN AMRO is counting on 35 per cent through AI or relocation abroad. The UWV estimates that about 13,000 people do this work across the sector. It already shows in the unemployment benefit figures. In January 2026 the number of CDD (customer due diligence) analysts on benefit had more than doubled from a year earlier, to about 260 people, and customer service staff are also coming onto benefit much faster than average. In its strategy of November 2025, ABN AMRO wrote that it wants to digitise processes end to end and integrate AI further.

The UWV and labour market experts add a caveat. AI implementation often still fails, and when giving notice of redundancies employers cite wage costs and uncertainty more often than technology. AI also creates work at first, such as checking its output and training staff. People who leave the sector rarely return: of those who moved from the financial sector onto unemployment benefit in 2024 and found work again, 22 per cent went back to that sector, against 44 per cent on average across all sectors. Staff are worried. In February 2026 the trade union CNV asked members at banks, insurers and pension administrators about their expectations, and 388 members responded. Of these, 63 per cent see the rapid development of AI as a threat to their own job, and a quarter do not expect to be working in the sector three years from now. Members say management presents AI as a tool, while they see cost cutting and job losses as the real aim. The survey is not representative: these are union members who chose to respond.

Internationally it fits what the supervisor and the consultants are seeing. More than 80 per cent of banks under ECB supervision name process automation as an important way to cut costs, the chair of the ECB’s Supervisory Board, Claudia Buch, said in Amsterdam in September 2026. In Capgemini’s survey of 1,100 leaders at banks and insurers, held in summer 2025, customer service, fraud detection, credit processing and onboarding new customers top the list of processes for which banks are deploying AI agents widely; 10 per cent of institutions already use agents widely. In its Global Banking Annual Review of October 2025, McKinsey calculates gross savings of up to 70 per cent in some cost categories and a net fall in total costs of 15 to 20 per cent, and expects competition to pass much of that on to customers.

How solid is this? The announcements are specific and the benefit figures per occupation are moving, but this is a single wave of restructuring at a handful of Dutch banks since late 2025, and the UWV does not yet see AI as the main cause. Whether the work really disappears because of AI will only become clear once the reorganisations are completed, between late 2026 and 2028.