Massachusetts review of DraftKings AI promotion model widens to neighbouring states

A spokesperson for the Maine Gambling Control Unit told a local television news team that the state has no laws or rules specifically governing artificial intelligence use by betting operators — a sentence that, read carefully, is a description of a gap rather than a reassurance.

Maine has two licensed mobile sportsbook operators. Both entered the market through partnerships with tribal nations, which hold exclusive rights under state law. That structure has so far insulated Maine from some of the commercial pressure visible in larger markets, but it has not insulated it from the underlying question: whether machine learning systems that analyse betting histories to determine promotional targeting constitute a form of harm the existing regulatory framework was designed to prevent.

That question arrived in Massachusetts first, after a New York Times investigation described DraftKings using algorithmic models to identify which customers were most likely to continue wagering after receiving a promotional offer. Former employees, cited in the reporting, said they were concerned the approach could worsen outcomes for people with gambling problems. DraftKings disputed the characterisation and said it complies with state regulations and does not market based on indicators of problem gaming.

Massachusetts opened a review. Maine said it was watching. Maryland's governor called for legislation that would prohibit betting apps from using algorithmic means to target users displaying problem gambling patterns. Three states, three different positions, none of them the same.

This is what a regulatory contagion looks like in its early phase. No enforcement has occurred in Maine. No rulemaking has been proposed. The Gambling Control Unit is, in its own description, monitoring — a word that occupies the space between awareness and action without committing to either. But the monitoring is happening now in part because the Massachusetts review gave it a frame, and the Maryland legislative push gave it a political shape, and the New York Times investigation gave it a public audience.

The consensus read is that AI targeting becomes a serious regulatory problem only when a major enforcement action lands. The consensus is probably wrong about the sequence. The more likely path runs through insurance: a state legislature that moves first, not because it has evidence of harm in its own market, but because it wants the liability settled before the harm is documented. Maryland's proposal fits that description exactly.

DraftKings has denied the specific conduct described. The denial is on the record. What is also on the record is that the company's promotional model has now attracted regulatory attention in at least three states simultaneously, and that the legal framework in each of those states was written before machine learning existed as a practical tool for operators. The regulators are not slow. The technology moved and the statutes did not.