New York moves to bar AI personalisation in sports betting promotions

Brian O'Dwyer has a problem he cannot solve alone. The chairman of the New York State Gaming Commission said it plainly after his panel voted unanimously to advance new restrictions on how the state's nine licensed sportsbooks use artificial intelligence: a bettor flagged as at-risk by one operator can walk to the next one, and the data does not follow them.

The rules O'Dwyer's commission advanced on Friday would prohibit mobile sports betting licensees from using AI, algorithmic profiling, machine learning, or generative systems to analyse customer data and personalise promotions, betting limits, odds, and product recommendations. A companion proposal would require operators to designate a specific person responsible for contacting patrons identified as potentially at risk for problem gambling.

The AI restriction is modelled on an existing New Jersey regulation, which stops short of banning the technology outright and instead draws a line at personalisation. That distinction matters commercially. The same systems that flag a compulsive gambler can also flag a high-value recreational customer. Operators have been careful to frame responsible gambling tools as complementary to their product rather than in tension with it. New York's commission is testing that framing directly.

O'Dwyer's language after the vote carried a specific kind of institutional threat: the industry should solve the data-sharing problem itself, or the commission would solve it for them. That phrasing is familiar from regulatory contexts where an agency has identified a structural failure but lacks the legal architecture to mandate a fix. The warning is real. The timeline is not yet specified.

The nine operators now on notice — Bally Bet, BetMGM, Caesars, DraftKings, Fanatics, FanDuel, Resorts World Bet, Rush Street Interactive, and theScore Bet — collectively represent the licensed mobile market in a state that generates more sports betting revenue than any other. The proposals still require additional public comment before a final vote, which gives operators a window to shape implementation or negotiate the data-sharing framework O'Dwyer is demanding.

The harder problem is the one he named. A cross-operator exclusion list requires operators to share customer data with each other and, implicitly, with the state. Every operator that has invested in proprietary risk-profiling systems has a commercial incentive to treat that data as confidential. The commission is asking them to treat it as infrastructure.

Where this sits in the broader regulatory picture is not incidental. The CFTC's moves to classify event contracts as federal swaps have consumed most of the attention on prediction market regulation. But New York's action is a reminder that the state layer of gambling oversight — the one the CFTC argues should be superseded — is still actively building its own standards, on its own schedule, in the largest mobile betting market in the country. Those two projects are not coordinated, and the gap between them is widening.