What "1,000 Jobs Replaced by AI" Actually Means

BNP Paribas Fortis says AI will absorb the work of 1,000 people by 2028. Read past the headline and the more instructive story appears: AI isn't the strategy here. Margin is. AI is just the most quotable lever.

Belgium's largest bank made news this week with a clean, alarming number: artificial intelligence and digitalisation could do the work of around 1,000 employees over the next three years. The framing wrote itself — robots take a thousand jobs. But the number rewards a closer reading, and what it reveals is a textbook example of how a bank actually converts AI into value: by subordinating it to a hard economic target and letting attrition do the quiet work.

Decode the number first

The thousand is not a thousand layoffs, and it is not all AI. By the bank's own account, AI today already performs the equivalent of roughly 270 people's work; the climb to 1,000 by 2028 is the genuine automation component. A second slice comes from re-insourcing tasks currently handled by external consultants — but running them in BNP Paribas group hubs in Lisbon and India rather than Belgium. So the cleaner statement is that around 1,000 functions leave Belgium, through a blend of automation and relocation, not that a thousand people are replaced by software.

And almost none of it requires firing anyone. Last year about 500 staff left and only around 330 were hired; the bank expects its Belgian headcount to fall by 100–150 a year through 2028, having already shrunk from roughly 13,500 in 2015 to about 10,000 today. CEO Michael Anseeuw is therefore able to say, accurately, that there is no plan for large-scale cuts to the workforce or branch network. Attrition is the change-management mechanism. AI determines how many of the leavers get replaced.

Where AI actually bites

The automation is aimed with precision at the highest-volume, lowest-judgement work. The clearest case is the call centre: bank-card queries alone generated 8.6 million customer contacts last year, and the bank wants to route 7 million interactions through its app's bot by 2028. The payoff is framed not as redundancy but as hiring avoided — by the bank's estimate, the bot work removes the need for roughly 290 new call-centre staff. AI is also being pointed at identity verification and fraud checks, two more high-volume, rules-heavy functions where it can sidestep hundreds of future hires.

This is exactly the pattern the broader industry data predicts. The boring, repetitive back-office task is where evidenced AI return lives — not the glamorous front-office use case. Fortis is simply executing it at a named scale.

The headline that wasn't reported

The cost story is only half the picture, and the quieter half is more strategic. The entire programme exists to hit a financial target: the French parent expects its Belgian arm to grow revenue by roughly 9% a year through 2028 — a rate the subsidiary has not recently achieved. Cutting cost-to-serve on commodity interactions is one side of that equation. The other is moving customers up the value chain. The bank expects around €5bn to flow into private banking and wealth management over the period, offers a dedicated private banker above €250,000 in assets, and will launch a digital investment platform in the first half of 2027 to compete with Belfius's ReBel and KBC's Bolero. It also wants 100,000 new customers a year, partly through its Bpost post-office partnership.

Put the pieces together and the "AI replaces 1,000 jobs" story is really a margin programme with three moving parts: automate the low-value interactions, relocate transactional work to group hubs, and re-segment customers toward higher-margin advice. AI is the connective tissue across all three — the same "one bank" integration logic the parent group preaches globally — not a standalone initiative.

The risk the number hides

There is an execution catch, and it is human, not technical. A bank can manage displacement through attrition only as long as its people and its social partners stay onside. Fortis already drew an unenthusiastic union response earlier this year when it transferred 580 customer-service employees to a Belgian arm of Accenture, and Anseeuw has been explicit about wanting to keep that dialogue open while pledging continued investment in training and diversity. The technology is the easy part; carrying the workforce and the works council through a multi-year reshaping is the hard part, and it is where programmes like this actually succeed or stall.

What BridgeUp is doing about it

The lesson generalises well beyond one Belgian bank. AI pays off when it is the servant of a defined economic strategy with hard numbers attached — a revenue target, a cost-to-serve curve, a customer-segmentation shift — and when natural attrition, not upheaval, absorbs the change. It disappoints when it is pursued as "adoption" for its own sake. The number that matters at Fortis is not 1,000 jobs; it is the 9% growth the whole exercise is built to deliver.

BridgeUp helps institutions design exactly that subordination: identifying the high-volume, low-judgement work where automation actually returns, sequencing it through attrition rather than rupture, redeploying freed capacity toward higher-margin services, and carrying the workforce through the transition so the plan survives contact with the people executing it. The tool is the easy half. The strategy — and the stewardship — is the work.

Sources: Belga News Agency, Brussels Times, La Libre, Knack/Trends, ICT Magazine (reporting on BNP Paribas Fortis's 2026–2028 "deep dive," released 1 June 2026); company statements by CEO Michael Anseeuw.