Spain’s Directorate General of Gambling (DGOJ) faces mounting scepticism from problem gambling experts over its soon-to-be-implemented machine learning algorithm, designed to detect risky player behaviours in the country’s online gambling market. The issue came to the fore during the recent Gaming in Spain Conference in Madrid, where DGOJ Director General Mikel Arana defended the regulator’s ongoing overhaul of compliance and governance in the sector.
DGOJ Pushes Forward with Regulatory Overhaul
Arana emphasised the significance of the DGOJ’s reforms, which have been in progress since 2020 under the Ministry of Consumer Affairs. He stated that three out of four central projects are now in the testing phase, with full implementation expected within six months. Among these initiatives, the introduction of a machine learning algorithm—using XGBoost technology and designed to detect 48 behavioural vulnerabilities—has been highlighted as a flagship achievement. The regulator also plans to prioritise the adoption of central deposit limits in the coming months.
Concerns Over Lack of Transparency and External Input
Despite the DGOJ’s confidence, industry experts have raised concerns about the development process of the new detection tool. Pedro Romero of BetBlocker criticised the lack of an open tender or collaboration with external partners during the algorithm’s design phase. Romero noted that greater dialogue and input from independent actors could have addressed key challenges in customer detection and intervention, warning that adopting models from other jurisdictions without adaptation may leave critical gaps.
Paula Murphy from Mindway AI echoed these concerns, highlighting the limited public consultation period and absence of external input into the algorithm’s parameters. Murphy questioned how operators would be expected to respond to the model’s findings, pointing out that the proposed system appeared to offer only binary outcomes for players, which may oversimplify the complexities of problem gambling.
Implications for Operators and the Industry
Operators and suppliers face uncertainty regarding the practical application of the new detection algorithm. Murphy stressed the importance of understanding the reasons behind customer flagging and tracking behavioural changes over time, rather than relying on a binary assessment. She also called for greater transparency about the training data and scope of the model, suggesting that the current approach may not leverage the full analytical capabilities already available in the market.
The lack of detailed guidance and transparency could present compliance challenges for operators, who may need to adapt their existing risk detection systems to accommodate the DGOJ’s model. The concerns raised underscore the need for ongoing dialogue between regulators, operators, and technology providers to ensure effective and nuanced player protection measures.
Key takeaways
- Spain’s DGOJ is moving forward with a new machine learning algorithm to detect problem gambling behaviours.
- Experts have criticised the lack of external input and transparency in the algorithm’s development.
- Operators may face challenges integrating the new system and understanding its outputs.
- Calls for greater dialogue and clarity highlight the need for collaborative regulatory approaches.