Dutch KSA Deploys Machine Learning to Detect High-Risk Gambling Behavior

The Dutch gambling authority, Kansspelautoriteit (KSA), is taking a significant technological leap forward to modernize player protection. By introducing a machine learning algorithm, the regulator aims to identify high-risk gambling behavior before it escalates into addiction. The system utilizes actual betting data to calculate an individual risk score. It goes beyond simple monetary amounts, analyzing complex patterns such as the frequency of wagers and the specific times of day when gambling occurs. A key feature is its focus on psychological components, monitoring how habits shift during winning or losing streaks.

This technological advancement is not a solo effort by the KSA but the result of close collaboration with academia. Researchers from the University of Amsterdam (UvA), including PhD candidate Charles de Leau and Professors Reinout Wiers (Psychology) and Johan Bollen (Computer Science), developed the model. The project was backed by the KSA’s Addiction Prevention Fund. The fact that the algorithm is available as open-source software marks a turning point in regulatory transparency. However, the KSA emphasizes that the technology is not a silver bullet. It serves as an independent tool to help licensed operators better fulfill their legal duty of care.

Numbers and facts

The model covers all forms of iGaming and is not restricted to online casinos. Inspiration for this system came directly from the Spanish regulator, Dirección General de Ordenación del Juego (DGOJ). The Spanish market is currently undergoing significant transformation, preparing for the introduction of cross-operator deposit limits. These will be tested during a six-month period starting September 25, 2026. The default limits are set at 700 Euro per day, 1,750 Euro per week, and 3,300 Euro over four weeks. While players can modify these limits, doing so requires specific procedures and additional safeguards.

"It offers no guarantees; it can only serve as a supplementary component of duty of care measures." - Spokesperson for Kansspelautoriteit, Regulator

In the Netherlands, the debate over such technical solutions is also fueled by criticism of existing strict deposit limits. Market participants fear that overly rigid restrictions could harm channelization rates. If players feel overly restricted by narrow financial limits, they might migrate to the unregulated black market. The KSA hopes that the algorithm will enable a more precise and less blanket form of protection, targeting individual risk rather than relying solely on fixed financial ceilings.

Background

The development of the algorithm shows that European regulators are increasingly focusing on cooperation. The Spanish DGOJ laid the groundwork with its Safe Play initiatives, aimed at better protecting minors and vulnerable groups. The goal is the earliest possible intervention. The KSA is building on these approaches, combining them with the expertise of the University of Amsterdam. The KSA Chairman noted in July 2026 that Dutch gambling requires better financial guardianship to maintain market integrity. The continuous development of systems like CRUKS, the central self-exclusion register, is part of this broader strategy.

Why it matters for German players

For German players, these developments in neighboring countries are highly relevant. While Germany already has very strict rules under the State Treaty on Gambling 2021 (GlüStV 2021), technical monitoring through algorithms could become a topic here as well. Currently, the German regulator (GGL) relies primarily on the LUGAS system. This monitors the cross-operator monthly deposit limit of 1,000 Euro and ensures the 1 Euro per spin limit for virtual slots is maintained. An intelligent algorithm based on the Dutch model could eventually be used to analyze player behavior more deeply within these limits.

German players currently benefit from the legal security provided by the GGL whitelist. Those playing at licensed providers are protected by the strict mandates of the GlüStV. However, experiences from the Netherlands show that purely financial limits are often criticized. A learning system that evaluates real gambling behavior could long-term become an alternative to the blanket 1,000 Euro limit, should authorities decide that individual risk assessment is more effective than a rigid cap for all citizens.

What it means for GGL-licensed casinos

German operators with a GGL license are already required to report extensive data to the authorities. If an algorithm similar to the KSA's were introduced in Germany, casino IT systems would require massive adjustments. The advantage for providers would be better legal protection: if a state-approved system certifies low risk, the danger of liability claims for breached duty of care decreases. At the same time, it increases the pressure to intervene immediately if the algorithm issues a warning. Compared to providers from Malta or Curacao, which often lack effective protection systems, this represents a clear mark of quality for the German licensed market.