New York is trying to answer a question the betting industry can no longer avoid: should artificial intelligence be allowed to identify what makes a customer more likely to wager and then use that knowledge to sell the next bet? The state’s emerging AI sports betting regulation draws a sharp distinction between technology used to protect customers and technology used to increase their wagering.
The New York State Gaming Commission unanimously advanced revised AI and responsible-gambling proposals on October 9, sending them into another public-comment period rather than making them final rules. The move comes as regulators elsewhere are asking similar questions; Massachusetts has already launched its own sportsbook AI review into how licensed operators use artificial intelligence and machine learning.
AI Sports Betting Regulation Is Moving Beyond Advertising
New York already regulates what sportsbooks can say to customers. Existing rules prohibit false or misleading advertising, promotions that encourage excessive wagering, and direct advice telling an individual patron to place a specific wager of a specific type or amount.
The AI proposal goes after a different layer of the relationship: the machinery deciding which customer sees which offer.
The Commission’s original responsible-gaming proposal sought to prevent gambling businesses from using AI-powered services to offer personalized promotions, suggested wagers or suggested wager amounts. The revised proposal advanced in October reportedly extends that concern to automated profiling and individualized decisions involving promotions, bonuses, odds, betting limits, messages and wagering recommendations.
That is a meaningful regulatory shift. The issue is no longer merely whether an advertisement is deceptive. It is whether the system choosing the advertisement knows too much about the person receiving it.
New York Is Separating Protection From Persuasion
The most interesting part of the proposal is that New York is not treating AI itself as the problem.
The state also wants sportsbooks to use automated systems to identify behavior that could indicate gambling harm. The March framework called for responsible-gaming leads, due-diligence reviews, intervention procedures and systems capable of flagging potentially at-risk customers.
That creates a deliberate regulatory line.
| Use of Technology | Regulatory Direction | Reason |
|---|---|---|
| Detect potentially harmful betting patterns | Encouraged or required | Helps identify customers who may need intervention |
| Provide responsible-gambling tools | Supported | Gives customers limits and assistance |
| Personalize promotions to an individual | Targeted for restriction | Can use behavioral data to stimulate wagering |
| Recommend specific bets or amounts | Targeted for restriction | Moves technology from information toward persuasion |
| Analyze customers for compliance | Necessary operational use | Supports regulatory and risk controls |
The principle is simple: use the data to reduce harm, not exploit vulnerability.
That distinction may eventually become more important than any specific definition of artificial intelligence. Operators have used algorithms, segmentation models and customer analytics long before generative AI became a boardroom obsession. Regulators therefore have to address automated decision-making broadly enough that replacing an “AI” system with an older predictive model does not become an easy workaround.
Personalization Is Where The Business Model Gets Complicated
Personalization is valuable because two customers do not behave the same way.
A sportsbook can analyze preferred sports, favorite bet types, deposit patterns, session frequency and promotional response. In ordinary digital commerce, using that information to make an offer more relevant is standard practice.
Gambling creates a harder problem because additional engagement is not always benign.
A retailer recommending another pair of shoes to a frequent buyer is not financially equivalent to a betting platform recognizing that a losing customer tends to deposit again after receiving an incentive. The product itself carries financial risk, and some patterns of repeated wagering can indicate harm rather than ordinary consumer enthusiasm.
That makes behavioral targeting unusually sensitive in betting.
The regulatory challenge is not to pretend personalization has no legitimate uses. A customer may genuinely prefer baseball markets over basketball, or straight bets over complex products. The difficult question is whether an operator should be permitted to combine that preference data with losses, deposits or indicators of distress to decide when and how aggressively to market another wager.
New York increasingly appears to be answering no.
Massachusetts Shows This Is Becoming A Wider Industry Issue
New York is not acting in isolation.
Massachusetts regulators recently asked licensed sportsbooks to explain how they use AI and machine learning in marketing, promotions, customer engagement and the evaluation of gambling behavior. Questions sent to operators included whether systems use losses, wagering frequency or other account behavior when making marketing decisions, and whether technology predicts future betting activity or chooses promotions likely to increase wagering.
The approaches are different but complementary.
Massachusetts is gathering information about what operators actually do. New York is moving toward defining practices it does not want them to do.
That distinction could matter nationally. State-by-state sports betting regulation often develops through imitation. Once one large jurisdiction creates a workable standard, other regulators can borrow its language rather than design a policy from scratch.
New York has particular influence because of the scale of its mobile sports betting market. A compliance system built for New York may eventually become easier for an operator to use across multiple states than maintaining sharply different AI marketing practices in every jurisdiction.
Compliance can become product design.
The Hard Part Will Be Defining The Boundary
The clean regulatory slogan is that AI should protect bettors rather than manipulate them. Implementing that principle will be harder.
Sportsbooks still need fraud detection, customer service automation, geolocation, responsible-gaming systems, risk management and normal marketing analytics. Regulators therefore cannot simply prohibit algorithms from touching customer data without disrupting legitimate functions.
Definitions will matter.
What counts as personalization? Does showing a baseball promotion to customers who regularly bet baseball violate the rule, or only tailoring it to a specific individual based on detailed behavioral profiling? Can an operator recommend a market if the customer explicitly requests assistance? How should odds or betting limits generated partly through automated systems be treated?
Industry comments during New York’s earlier consultation already pushed for distinctions between technology intended to exploit gambling behavior and tools intended to educate or assist customers. The New York Racing Association, for example, argued that AI-powered analytical tools for horseplayers should not automatically be treated the same as AI used to pressure vulnerable consumers.
Those edge cases are why the public-comment process matters.
The Next Fight Is Over Who Controls The Algorithm
Sportsbook regulation used to focus heavily on visible things: advertisements, bonus language, age restrictions and licensing.
AI moves regulation inside the decision engine.
A promotion can look perfectly ordinary on the screen while the system behind it has analyzed thousands of signals to determine that one particular customer is unusually likely to respond after a losing session. That makes oversight more difficult because regulators need to understand not just what the customer saw, but why that customer saw it.
New York’s proposal therefore points toward a more demanding compliance environment. Operators may need clearer records of which data feeds marketing systems, how automated decisions are made, which customers are excluded from promotional targeting and how responsible-gambling safeguards override commercial objectives.
The rule is still a proposal, not settled law. The revised measures are going through public comment before they can return to the Commission for final consideration.
But the direction is already important.
AI sports betting regulation is becoming less about whether sportsbooks may use artificial intelligence at all and more about which side of the customer relationship AI is allowed to serve. New York is proposing that technology can watch for danger, assist with protection and improve compliance, but it should not quietly learn a bettor’s weaknesses and turn them into a marketing opportunity.
