Beyond the Black Box: A Benchmarking Framework for AI Player Risk Detection in Gambling

Session Title

AI & Technology: Industry Frameworks

Presentation Type

Paper Presentation

Start Date

26-5-2026 12:00 AM

Abstract

Artificial intelligence is now the engine behind gambling harm prevention, yet the industry lacks a "yardstick" to measure its success. Without standardized evaluation, it is impossible to determine which AI models actually protect players and which simply provide a false sense of security. This paper moves past the theoretical by proposing a domain-specific conceptual framework for benchmarking AI-enabled risk detection. By defining standardized datasets, specific detection tasks, and uniform performance metrics, this framework allows for the first objective: longitudinal comparison of player risk systems. We outline how this structured approach empowers regulators, operators, and developers to validate AI effectiveness, ensuring that innovation in harm prevention is both transparent and evidence-based.

Author Bios

Simo Dragicevic has been a technology entrepreneur in the gambling industry since 2010, as founder of BetBuddy, a pioneer in Artificial Intelligence and gambling, which was acquired by Playtech Plc in 2017. He is also the Co-founder of AiR Hub, the industry's first, dedicated AI research lab established at UNLV's International Gaming Institute.The mission of AiR Hub is to advance the emerging discipline of AI within the gambling sector and create the industry’s ecosystem for collaborative AI research. Its flagship State of AI in Gaming annual report is the first comprehensive, data-informed overview of AI in the global gambling sector.Previously, Simo was Managing Director and Head of AI at Playtech and has also served on the boards of the Great Britain regulator, the Gambling Commission, and the Responsible Gambling Council. Prior to founding BetBuddy Simo was Director of Major Programmes at Barclays Plc and began his career at Accenture.

Share

COinS
 
May 26th, 12:00 AM

Beyond the Black Box: A Benchmarking Framework for AI Player Risk Detection in Gambling

Artificial intelligence is now the engine behind gambling harm prevention, yet the industry lacks a "yardstick" to measure its success. Without standardized evaluation, it is impossible to determine which AI models actually protect players and which simply provide a false sense of security. This paper moves past the theoretical by proposing a domain-specific conceptual framework for benchmarking AI-enabled risk detection. By defining standardized datasets, specific detection tasks, and uniform performance metrics, this framework allows for the first objective: longitudinal comparison of player risk systems. We outline how this structured approach empowers regulators, operators, and developers to validate AI effectiveness, ensuring that innovation in harm prevention is both transparent and evidence-based.