France's regulator has built a tool that lets operators measure whether their own player protection systems are actually working. The findings should worry the entire industry: excessive players may account for 60% of account-based gambling revenue.
The Autorité Nationale des Jeux has rolled out a player-risk algorithm that is likely to reshape how France's licensed gambling operators approach their responsible gambling obligations. The tool is being offered as an optional benchmark that operators can use alongside their existing detection systems, but the data it has already generated about the scale of excessive play in the French market makes its findings difficult to ignore regardless of whether individual operators choose to adopt it.
The algorithm itself is a methodologically serious piece of work. It draws on 23 risk indicators spanning financial movements, play frequency, use of moderation tools, player activity patterns and previous behavioural history. Players are sorted into four categories: recreational, moderate risk, excessive and manifestly excessive. Critically, the model was validated against the Canadian Problem Gambling Index, a recognised international screening instrument, under the supervision of a scientific committee. That validation process gives the algorithm a credibility that purely operator-built risk models often lack, and it provides the ANJ with an authoritative reference point for assessing whether the industry's own detection efforts are matching the scale of the underlying problem.
The detection numbers tell a story of both genuine progress and persistent shortfall. The ANJ has acknowledged that operators have made significant strides in identifying excessive or pathological gamblers, with the number of players identified rising from 31,000 in 2024 to 89,000 in 2025. That increase reflects real investment in behavioural analytics, safer gambling staff and intervention workflows across the licensed market. For an industry that has historically struggled to demonstrate meaningful improvement in this area, tripling identification in a single year is a substantial result.
The problem is that the ANJ's own algorithm suggests the industry is still seeing only a fraction of the excessive players actually present in the market. The regulator's model identified 600,000 players with a strong probability of being excessive during the second half of 2025, equal to 8.7% of account-based players. Approximately 300,000 of those were classed as manifestly excessive, the most serious category in the algorithm's classification framework. The gap between 89,000 players identified by operators and 600,000 flagged by the ANJ algorithm is the central finding that will drive the next phase of regulatory engagement.
The commercial dimension of the data is what is likely to focus boardroom attention. The ANJ estimates that excessive players generated €1.2 billion in GGR across the account-based market covered by the algorithm, representing 60% of GGR within that specific scope. That share has been increasing since 2023. The implication is uncomfortable: a substantial majority of revenue in the regulated French online market is being generated by players the regulator's own model considers to be gambling excessively. Operators have always understood that high-value players carry disproportionate revenue significance, but having that observation quantified at this scale and connected explicitly to excessive gambling categorisation changes the regulatory conversation in fundamental ways.
The ANJ has been clear that the algorithm is not intended to replace prevalence studies or to produce medical diagnoses. Its purpose is to provide both operators and the regulator with a shared reference point for prevention work. France's existing responsible gambling framework already requires operators to identify customers showing signs of excessive or pathological play and to support them in bringing their gambling back under control. The regulator's practical guide stresses that operators are not healthcare professionals and are not expected to diagnose players. Their role is to detect characterised levels of risk from observable behaviour, and the new algorithm gives them a benchmark against which to assess whether their detection systems are operating at the right level.
The intervention framework the ANJ recommends is graduated. General information for lower-risk cases. Personalised messages and observation for moderate risk. Direct contact including telephone calls for the most serious cases. Available tools include deposit and loss limits, self-exclusion, removal from commercial communications, referral to support services and, in severe cases, voluntary gambling bans. The guide warns specifically against relying only on big player lists, noting that high spenders are more likely to include excessive gamblers but are not automatically problem gamblers. Effective detection requires combining financial data with time spent gambling, betting patterns, moderation tool behaviour and customer service interactions.
The illegal market dimension complicates the picture considerably. The French online gaming association AFJEL has warned that illegal gambling sites attracted 5.4 million French players in 2025 compared with 3.5 million on the regulated market. Illegal GGR is estimated at €2 billion, and 62% of players using illegal sites are reported to show excessive or pathological gambling behaviour. That figure is far higher than the 8.7% identified by the ANJ algorithm in the licensed market, reflecting both the absence of safeguards on illegal sites and the structural likelihood that vulnerable players who have been identified and constrained by licensed operators will migrate to unregulated alternatives.
Of all the data points the ANJ has produced, the finding that excessive players generate 60% of account-based GGR is the most consequential. That figure transforms the responsible gambling debate from an abstract question of consumer protection into a direct conversation about the commercial model of regulated French gambling. If the majority of revenue across the licensed online market is being generated by players the regulator categorises as gambling excessively, then any meaningful effort to reduce excessive play has substantial commercial implications for operators. Conversely, an industry that depends on excessive play for the majority of its revenue faces a structural credibility problem with the regulator, the political establishment and the public. The ANJ has not yet drawn explicit conclusions from this data, but it has put a number on the table that will frame every future conversation about regulatory tightening in the French market.
The contrast between 89,000 players identified by operators and 600,000 flagged by the ANJ algorithm cuts two ways. From a regulatory perspective, it provides clear evidence that the industry's own detection systems are missing the majority of excessive players, supporting any future case for stricter requirements or more prescriptive interventions. From an operator perspective, it provides a roadmap for improvement that could meaningfully reduce regulatory risk if acted upon proactively. The operators that adopt the ANJ algorithm, demonstrate alignment between their own detection results and the regulator's benchmark and document genuine intervention activity against the players newly identified will have a substantially stronger position in their next compliance review than those that continue to rely on internal systems that the regulator can demonstrably show are missing most of the relevant cases.
The 62% excessive gambling rate among players on illegal French sites compared with 8.7% on licensed platforms is the most important number in this story. It reflects the reality that France's regulatory framework is doing genuinely important work in protecting players who remain within the licensed market, but that work is being undermined by the structural availability of alternatives that face none of the same constraints. Every excessive player who is successfully identified and constrained by a licensed operator becomes a candidate for migration to an unregulated platform where their behaviour will not be detected or addressed. The ANJ's investment in better detection within the licensed market needs to be paired with equivalent investment in restricting access to and disrupting the unlicensed market, or the licensed sector's improvements will simply accelerate the migration of vulnerable players toward worse outcomes. France has done excellent work on one side of this equation. The other side urgently needs comparable attention.
