Massachusetts Seeks Details on AI Sportsbook Marketing Systems
Sportsbook promotions can be shaped by a customer’s betting history, losses and deposits. Massachusetts regulators are asking seven licensed mobile operators to map the systems behind those decisions, with written responses due November 3. The inquiry puts a practical question at the center: can an operator explain why a particular person received a particular offer? That question matters because a promotion can look routine to the customer while being the result of a series of automated choices made behind the scenes. The commission is asking operators to make those choices legible.
What the commission wants to see
An October 9 questionnaire from Massachusetts Gaming Commission Executive Director Dean Serpa asks whether operators use artificial intelligence, machine learning, algorithms or automated decision systems in marketing, promotions and customer engagement. Companies must identify the tools, describe their purposes and say whether they built them or obtained them from a third party. Those distinctions can help show where responsibility sits: an operator may rely on a vendor’s product, but it still needs to explain how that product is used with its customers. The questionnaire seeks a practical inventory, not just a broad assurance that technology is involved.
The commission also wants a view of how these systems work at the customer level. Operators are asked whether decisions apply to individuals or customer segments, what information feeds those decisions, and how companies determine who receives which offers and when. A segment based on shared characteristics raises different questions from a decision tailored to one person’s account activity, even if both approaches rely on automated tools. Timing matters, too: knowing what triggers an offer can help explain whether a campaign is based on general engagement or on a customer’s recent behavior. The answers should give regulators a clearer picture of how data becomes a marketing decision.
That focus extends to betting and account activity. The questionnaire asks whether recent wagers, losses, deposits or other customer behavior can trigger marketing. It also asks whether heavier losses, more frequent betting or signs of potentially harmful gambling increase someone’s likelihood of receiving a promotion. These are pointed questions because the same information that helps a company understand customer activity could also shape who sees an incentive. The commission is asking operators to describe that connection directly, rather than leaving it buried in general descriptions of their marketing systems.
If those indicators don’t raise the chance of a customer getting an offer, operators must explain the controls that prevent them from doing so. The questions also cover behavioral targeting, responsible gambling, how decisions are made and where human oversight enters the process. A written policy can describe a safeguard, but the commission’s questions also invite operators to explain how it applies to actual decisions. That’s where the details matter: a control is only useful to regulators if they can understand what it’s intended to prevent and how people oversee the system. This is the part most people overlook: a model’s purpose and its practical effect aren’t always the same thing.
Why AI-driven targeting is under scrutiny
The commission announced its review at a September 24 meeting. Chair Jordan Maynard said the MGC would assess each licensed sportsbook’s AI use and its effects on Massachusetts customers. The move followed a New York Times report alleging DraftKings used AI and machine learning to target bettors considered most likely to lose. That report provides the backdrop for the inquiry, but it doesn’t establish what any of the seven operators in Massachusetts do. The commission’s questionnaire is the mechanism it’s using to gather information from each licensee.
The questionnaire doesn’t establish that any operator has used its systems to target customers in those ways. It asks each company to describe its practices so the commission can understand how promotional decisions are made and what safeguards are in place. That distinction matters: asking whether a system can act on betting patterns isn’t a finding that it has done so, or that any particular offer was driven by losses. The questions also touch on sportsbook promotions tied to predicted losses, including the role that betting patterns and account activity may play. The numbers here are underwhelming in one respect: the questionnaire can reveal what operators say their systems do, but the responses alone may not settle how those systems behave in practice.
Written answers may shape the next phase
Massachusetts is gathering operator disclosures, while New York regulators have proposed restrictions on using AI to personalize promotions or recommend bets, according to RG.org. Sports Betting Dime has reported that the MGC also plans to interview sportsbook representatives about AI in customer acquisition, promotions and responsible gambling. These steps show regulators examining the issue through both written descriptions and direct discussion. The two forms of inquiry can add different kinds of detail: written answers create a record of what companies report, while interviews can give officials an opportunity to ask how those policies work in practice.
The commission’s review of written responses and interviews could help determine its next steps as it develops policy on AI. For now, each of the seven operators faces the same November 3 deadline. Their answers will show how clearly they can account for the data, systems and safeguards behind customer offers. If you’re working in this space, the questions point to a basic test of accountability: can a company trace a promotion from the information it used to the decision it made, and explain who checks that process? That’s a useful test even before the commission decides what policy should follow. A paper trail, at least.
The post AI Questions Put Massachusetts Sportsbooks Under Review appeared first on ReadWrite.
