TL;DR

Roughly 241,000 drivers across Britain and the EU are behind a compensation claim filed in Amsterdam, arguing that Uber’s personalised pay-setting model profiles them unlawfully under GDPR. The European Trade Union Confederation calls it the first collective action of its type. Damages could reach into the billions.

What the drivers say the model does

The complaint targets a system that prices each trip individually using what it has learned about the driver being offered it. Two men taking a break together in Tottenham each got an offer for the very same job, at £23 and £27; the lower-paid driver had recently taken several cheap fares. Drivers also report smaller offers for return legs, on the reasoning that nobody wants to drive home empty.

Uber rejects this squarely. It says driver-level acceptance history plays no part in what a trip pays, attributing variation instead to surge pricing, promotions, satellite positioning and live testing, and pointing out that drivers see the fare and destination before accepting. “We categorically reject the allegations,” a spokesperson said.

The claim runs wider than pay. It alleges driver data was used unlawfully to train Uber’s models, and seeks an injunction as well as damages. Britain is where the mechanism has run longest — since 2023, the claimants say, costing UK drivers around £5,000 a year each. Dutch drivers only came under it this year.

Why this one has teeth

The case is led by Worker Info Exchange, whose founder James Farrar won the Supreme Court ruling that made Uber drivers workers in UK law. That is not a group that files speculatively. It arrives a month after the Netherlands’ privacy regulator hit Uber with a £708m (€825m) penalty over automated account deactivations, now under appeal, and alongside 2025 Oxford research finding earnings fell materially once dynamic pricing arrived. Uber disputes that study’s data.

Looking forward

Any British employer running AI over its own workforce should read the pleadings rather than the headlines. The legal theory is not that the outcome was unfair — it is that automated profiling took a decision about someone’s livelihood without a lawful basis or adequate explanation. That test travels well beyond ride-hailing, into rota-setting, performance scoring and pay banding.

It also lands while English law is being told it already answers who pays when AI causes harm. Disclosure is the immediate exposure: a firm that cannot explain what its model weighed will struggle to defend the decision it produced.