Spanish regulator to add indicators to ambitious AI gambling surveillance system
The DGOJ will use AI to identify patterns linked to risky behaviour.
Spain.- The Spanish gambling regulator, the DGOJ, has unveiled plans to incorporate over 60 behavioural and transactional indicators into its AI-driven surveillance system. The initiative is designed to track online gambling activity in real time and pinpoint patterns linked to risky or problematic behaviour.
The aim of the project is to detect early warning signs in order to facilitate personalised interventions by operators. Promoted by the Ministry of Consumer Affairs, it stems from the 2023 Royal Decree on Safer Gambling Environments, which mandated the DGOJ to develop tools that mitigate gambling-related harm.
When first introduced, the DGOJ described the system as a potential benchmark for Europe, with government officials touting it as the most advanced gambling oversight model within the European Union.
The system will be powered by the machine learning algorithm XGBoost. It will assess variables, such as session frequency and intensity, betting behaviour following wins or losses, and patterns in deposits and withdrawals. Additional metrics include cash-out habits, consecutive days of play, and preferences for live betting – an area the regulator sees as particularly high-risk due to its fast pace and impulsive nature.
To refine the system’s design, the DGOJ has launched a public consultation open until September 25. Stakeholders including operators, tech firms, researchers, and citizens are invited to contribute feedback on the model’s technical and legal dimensions.
The project has faced scrutiny over its limited coverage and uncertain rollout schedule. The AI system will monitor sports betting, poker and casino casino games but will exclude the lottery products of entities like SELAE and ONCE. which remain the most popular form of gambling in Spain. Moreover, while the Royal Decree allows for a two-year development period, the DGOJ has yet to announce a timeline for beta testing or to clarify how risk scores will be harmonized across platforms.
