AI gambling promotions are among the subjects of new questions from US lawmakers to six betting and prediction-market businesses. In a 22 October announcement, Senator Richard Blumenthal and Representatives Paul Tonko and Valerie Foushee identified BetMGM, Caesars, DraftKings, Fanatics, Kalshi and Polymarket as recipients.
Questions about VIP treatment
The letters request replies by 5 November 2026 about customer losses, VIP admission, incentives and whether AI tracks or targets bettors with offers. The lawmakers express concerns about gambling harm; their allegations are not findings of misconduct by the recipients.
The separate Massachusetts AI questionnaire also seeks information about automated promotional decisions. These are distinct requests from different public bodies.
Understanding the incentive behind an offer
Our assessment is that operators could gain a clearer picture by separating admission to a VIP programme from the decision to send each subsequent offer. Those decisions may sit in different products or teams. Explaining how membership works would not necessarily explain why a particular customer received a particular incentive.
A useful starting point would be the objective assigned to a model. A system trained to predict a response to a promotion answers a different question from one designed to identify potential gambling harm. Operators could examine how those outputs interact when both are available, and who decides what happens when they point in different directions.
The same distinction matters when measuring a campaign. An increase in activity could look commercially successful while leaving unanswered questions about the customers driving that increase. Reviewing aggregate results alongside the treatment of customers flagged for further attention could expose gaps that a single conversion metric would miss.
For suppliers, this creates a practical product question: can an operator inspect the objective, inputs and controls behind an offer without reconstructing the process manually? Useful reporting would connect those elements to the relevant decision, while making clear where a human changed the result.