Kindred Group Launches AI-Powered Harm Detection System Across European Brands

Kindred Group announced today the rollout of PS-EDS 3.0, the third generation of its Player Safety Early Detection System, across its portfolio of European-facing brands including Unibet, Maria Casino, and 32Red. The platform, developed in partnership with researchers at Linköping University, now processes more than 200 behavioural markers in real time and leverages federated learning techniques to improve detection accuracy without centralising sensitive player data across jurisdictions.

According to internal testing data shared by the operator, the upgraded system identifies escalating risk behaviours an average of 48 hours earlier than its predecessor, giving customer interaction teams a wider window for proactive outreach. Kindred reported that pilot deployments in Sweden and the Netherlands during Q3 produced a 22% increase in meaningful player interactions and a 14% reduction in reported harm indicators among flagged accounts. The operator confirmed that its share of revenue derived from high-risk play dropped to 2.9% in the most recent quarterly disclosure, continuing a multi-year downward trend.

Chief Executive Nils Andén said the investment reflected both regulatory momentum across Europe and Kindred's longstanding commitment to its 'Journey Towards Zero' strategy. The company noted that PS-EDS 3.0 has been designed to integrate with incoming requirements under the Dutch KSA's updated duty-of-care rules and Germany's GGL safer gambling framework, including automated documentation trails that satisfy supervisory audit standards in both jurisdictions.

Industry analysts suggested the launch raises the technical baseline for responsible gambling tooling across tier-one European operators, several of whom have been under pressure to demonstrate measurable harm-reduction outcomes rather than process-based compliance. Kindred confirmed that a white paper detailing the system's methodology and anonymised performance data will be published in early 2027, with portions of the framework to be made available to academic researchers under a controlled access agreement.