DraftKings is using AI to behaviorally target chronic gamblers

The practice of using data analysis to identify and target specific customer segments has deep roots in mainstream business, predating artificial intelligence by decades. Retailers have long used purchase history, location data, and demographic information to decide which customers receive which offers. Insurance companies built actuarial models to assess risk profiles. Banks developed credit scoring systems. What's different now is the scale, speed, and sophistication with which patterns can be detected and acted upon. Machine learning systems can process vastly larger datasets and spot correlations that humans might miss, leading to more granular segmentation of audiences into smaller, more specific behavioral groups. When applied to any product or service, AI based targeting works by identifying users who exhibit certain characteristics and then increasing exposure or engagement with that group. The system observes behavioral patterns like frequency of use, spending volume, session duration, or the timing of interactions. It learns which of these patterns tend to predict future behavior. Once a model is trained, it can flag new users as matching a particular profile and automatically adjust what content, offers, or messaging they see. This happens at scale and often in real time, without human review of individual cases. The appeal to any business is obvious: resources flow toward customers most likely to engage further. The gambling industry, specifically sports betting and daily fantasy sports, operates within a regulatory and business environment where customer acquisition and retention are central to the business model. Like other entertainment platforms, betting sites use conventional marketing data to understand their users. They know who bets frequently, who bets large amounts, who bets at certain times, and how betting patterns shift. Understanding these patterns allows platforms to optimize their operations. The tension arises because the data that predicts future betting behavior is the same data that can identify users whose betting patterns may reflect dependence or loss of control, regardless of how that identification is used or disclosed. The broader context here involves the capabilities and incentive structures of AI systems. Machine learning excels at prediction. A well trained model can identify patterns in user behavior with high accuracy long before those patterns become obvious to the user themselves or to external observers. An AI system trained on historical data can recognize the profile of a user at risk of problematic engagement. Whether and how that information is used depends on the policies and choices made by the company deploying the system. Some platforms might use such insights to implement safeguards. Others might use similar data to optimize marketing spend by targeting high engagement users, which could include users exhibiting signs of problematic use patterns. Behavioral targeting raises questions abo