Browser Dating: The Art Project That Exposes Dating's Data Dilemma

Belgian artist Dries Depoorter has created Browser Dating, a platform that matches users based on their browsing history rather than profile photos. Fewer than 1,000 people have signed up since launch. The dismal uptake reveals more about consumer tolerance for data exposure than any privacy survey ever could.

The surveillance artist goes after dating

Depoorter isn't trying to build the next unicorn. His previous work includes software that cross-references LinkedIn profile photos with security camera footage to identify people in public spaces, and a tool that catches Flemish politicians scrolling their phones during parliamentary sessions. Browser Dating sits in this same tradition: art that functions as working technology, designed to make you uncomfortable about what's already happening.

The mechanics are straightforward. Users install a browser extension that analyses their last 5,000 searches and visited URLs. The platform claims to process this data locally, then uses what Depoorter describes as AI to match users based on browsing patterns rather than selfies or stated preferences. The entire setup is presented as a more "honest" approach to compatibility: your search history doesn't lie about your interests the way a curated bio might.

Except browser history doesn't just reveal interests. It reveals late-night health anxiety, financial desperation, political rabbit holes, and every half-formed curiosity you've ever had about subjects you'd never discuss on a first date. Context collapses. A single search for symptoms of depression could reflect genuine struggle, research for a friend, or background for a novel.

This is performance art disguised as product development, and the dating industry should be grateful it's being asked these questions by an artist rather than a regulator with subpoena power.

The data contradiction nobody's explaining

The central technical claim doesn't withstand scrutiny. Depoorter states that browsing data is "processed locally and not shared with third parties". If true, this would mean the matching algorithm runs entirely on-device, comparing your local data fingerprint against others without uploading raw histories to a central server. Possible, but technically complex and computationally expensive for a solo art project.

More likely, some form of derived data, hashed patterns, categorical tags, or anonymised vectors, is being transmitted for matching purposes. That's not the same as uploading raw URLs, but it's also not "local processing" in the way most users would understand the term. The project's documentation doesn't clarify this, and Depoorter hasn't provided technical specifics about the architecture.

This matters because the dating industry has spent years finessing the language around data collection. Match Group's suite of apps collects device data, location history, and in-app behaviour. Bumble (BMBL) uses photo verification and has experimented with video prompts. Grindr (GRND) has faced repeated criticism for sharing HIV status and precise location data with advertising partners. All position themselves as privacy-conscious whilst building business models that require continuous behavioural tracking.

Browser Dating collapses that careful ambiguity. Either you're uploading one of the most intimate datasets you generate, or you're not, and if you're not, the matching can't work as advertised.

What the uptake figures actually mean

A thousand users for an experimental art project isn't failure, it's proof of concept for the thesis. Dating apps routinely claim they want to facilitate "authentic" connections, yet users are demonstrably unwilling to provide the data that would enable truly predictive matching. The gap between what platforms say they want to build and what users will actually tolerate is the story here.

Consider the contrast with mainstream adoption patterns. Tinder reached 50 million users by making dating frictionless: swipe on a photo, match if mutual, start chatting. Hinge rebuilt its entire brand around "designed to be deleted", emphasising intentionality without requiring invasive data. Feeld carved out a niche for non-monogamous users by creating space for preferences the mainstream platforms ignored, not by demanding access to browser histories.

The most successful dating products of the past decade have added friction selectively, video verification to reduce catfishing, prompts to improve conversation starters, whilst keeping core functionality simple. Browser Dating does the opposite. It asks for maximum data exposure upfront, before you've seen a single match or had a conversation. That's not how consumer trust works, and it's certainly not how dating works.

The regulatory context arriving faster than founders realise

Depoorter's project lands as European regulators are tightening enforcement around data minimisation and purpose limitation. The EU Digital Services Act (DSA) requires platforms to demonstrate that data collection is necessary and proportionate. The UK Online Safety Act (OSA) will mandate age verification and impose duty of care obligations that extend to how platforms use personal data.

Browser histories would be classified as special category data under GDPR in many scenarios: health searches, political affiliation, sexual orientation. Collecting and processing them for matchmaking purposes would require explicit consent, robust security measures, and a legitimate basis that survives regulatory challenge. Depoorter's positioning as art likely exempts him from enforcement, but a commercial operator attempting the same model would face immediate scrutiny.

Dating operators should take note. The direction of travel is toward less permissive data practices, not more. The ICO has already investigated dating apps for insufficient age verification and poor data retention policies. Browser Dating might be a provocation, but the questions it raises, how much data is too much, who benefits from hyper-personalisation, what happens when prediction crosses into surveillance, are the ones compliance teams will be answering under oath.

The fact that Depoorter built this to make a point about surveillance doesn't mean someone else won't try to build it for profit. The real test will be whether the industry can articulate where the line is before regulators draw it for them.