France’s gambling regulator has released an algorithm suggesting that 60% of the country’s gross gaming revenue comes from problem gamblers. The finding raises some uncomfortable questions about operator detection systems and what effective safer gambling actually looks like in practice.

The Algorithm’s Troubling Baseline

The ANJ (Autorité Nationale des Jeux) developed the tool using data from licensed online operators plus FDJ and PMU figures. It identifies at-risk players across four risk categories using 23 distinct indicators, validated against the Canadian Problem Gambling Index and reviewed by expert committees.

The results are stark. In the second half of 2025 alone, the algorithm flagged 600,000 players as very likely problem gamblers. That breaks down to 300,000 clearly excessive players and another 300,000 highly risky cases. What really matters, though: those €1.2 billion in problem gambling revenue represents a genuine revenue concentration issue.

A Detection Gap Nobody Expected

Here’s where it gets interesting. Operators reported identifying 89,000 problem players in 2025, up from 31,000 the year before. Looks decent on paper. But the ANJ’s independent analysis suggests they’re missing the vast majority of at-risk individuals.

The regulator acknowledged that market expansion explains some of the growth in problem player numbers. But it noted the rate of increase outpaces total player growth. That suggests something more systematic is happening.

An Optional Benchmark Tool

Rather than mandate compliance with a single detection method, the ANJ has positioned its algorithm as a benchmarking tool operators can adopt voluntarily alongside existing systems. It gives operators flexibility while providing objective measurement of their safer gambling efforts.

The regulator will itself use the algorithm to track clearly excessive players and monitor emerging risk cases. ANJ chair Isabelle Falque-Pierrotin called the release a decisive regulatory step. She also flagged, though, that detection needs to happen at retail points of sale as well as online.

What This Means

The 60% finding isn’t a claim about the true scale of problem gambling. It’s a statement about revenue concentration among identified at-risk players. That distinction matters. But it does underline something real: are current detection mechanisms fit for purpose?

Operators now have a credible external benchmark to test their own systems against. That’s useful. Whether they’ll meaningfully adopt it? We’ll see.

What the team thinks

Philippa Ashworth says:

Baz has surfaced a genuinely important conversation here, though I’d argue the real story isn’t the algorithm itself but what it exposes about how we measure operator compliance versus actual harm reduction. The €1.2bn figure is certainly attention-grabbing, but we need to distinguish between revenue correlation and causation, because operators with robust detection systems may naturally show higher problem gambler percentages simply because they’re identifying risk more effectively than their competitors. What France’s regulator has given us isn’t a smoking gun so much as a diagnostic tool, and the industry’s response to these findings, particularly whether operators invest in better early intervention rather than just better at-risk player identification, will tell us far more about the state of safer gambling than any single algorithm ever could.