New York aims to restrict sports betting licensees’ use of AI

Dig Deeper

The Backstory

AI scrutiny moves from guidance to potential limits

New York’s proposal marks a sharper turn in the gaming industry’s debate over artificial intelligence: from whether operators should use AI responsibly to whether certain uses should be restricted outright. The state’s gaming regulator is targeting the commercial core of online sports betting, where customer data, automated segmentation and personalized incentives can shape what players see, when they see it and how often they return.

The move follows growing concern that the same tools used to improve customer service, detect fraud and identify risky behavior can also be used to increase engagement among vulnerable players. Sportsbooks have long relied on data analytics to tailor bonuses and messaging, but generative AI, machine learning and algorithmic profiling have expanded the speed and precision of those efforts. Regulators are now examining whether that capability creates risks that existing responsible gambling rules were not designed to address.

That tension is evident in how regulators outside New York have approached the issue. The Malta Gaming Authority recently published a voluntary AI Gaming Charter for responsible AI use, laying out best-practice guidance rather than new binding rules. Malta’s framework recognizes AI’s role across responsible gambling, anti-money laundering, fraud prevention, customer service, payments and game development. But it also points to the same weaknesses that appear to be driving New York’s rulemaking: gaps in governance, risk assessment, transparency, bias testing and safeguards for vulnerable players.

Regulators are trying to separate protection from persuasion

The regulatory challenge is that AI is not inherently harmful in gambling. Some of its most widely promoted uses are protective. Operators and suppliers use automated monitoring to spot changes in play, detect unusual deposits, identify markers of harm and trigger interventions. Those tools can help companies comply with safer gambling obligations and act faster than manual review systems.

The question is whether the same behavioral data can be used for conflicting purposes. If a system can identify a customer who may be at risk, it can also identify a customer likely to respond to a bonus, a push notification or a suggested bet. New York’s proposed restrictions appear designed to prevent operators from using advanced profiling to personalize promotions, bonuses, odds, betting limits, messages and wagering recommendations in ways that could increase spend or deepen risky play.

Industry technology providers have leaned into the protective side of the equation. Mindway AI and GamScore recently announced a strategic partnership focused on safer gambling frameworks, combining user-controlled financial health tracking with AI-based behavioral monitoring. The partnership reflects a broader industry push to present AI as a tool for sustainability, giving players more control over their data while providing operators with deeper risk-detection insights.

That type of application is unlikely to disappear under stricter regulatory regimes. In fact, New York is advancing its AI proposal alongside a separate measure requiring sports betting licensees to identify patrons who may be at risk for problem gambling. The two tracks show where regulators may draw the line: using AI to detect and reduce harm may be encouraged or required, while using AI to increase personalized engagement may face heavier limits.

Malta’s voluntary model contrasts with New York’s enforcement posture

Malta’s AI charter reflects a European-style governance approach shaped by the EU AI Act and the General Data Protection Regulation. It calls for human oversight, transparency, fairness, security, accountability and ongoing monitoring, while urging operators to build internal structures for risk assessment and incident response. The charter is voluntary, but it signals that regulators expect operators to document how AI systems work and how risks are controlled.

New York is moving closer to a rule-based approach. Its proposal would apply to sportsbooks, affiliates and vendors, an important distinction because much of the modern betting ecosystem sits outside the licensed operator’s direct marketing department. Affiliates drive acquisition, suppliers power personalization engines and third-party vendors often manage customer engagement tools. Extending restrictions across that chain would make compliance more complex and could force operators to revisit vendor contracts, consent flows and marketing systems.

The contrast also shows how fragmented AI regulation in gaming may become. A Malta-licensed operator could be encouraged to adopt voluntary governance principles, while a New York-licensed sportsbook could be barred from using comparable systems for certain customer-facing commercial purposes unless specific consent conditions are met. For global operators, the operational burden is not just complying with one AI policy, but building systems that can adapt to different jurisdictions’ definitions of profiling, personalization and acceptable use.

Player data has become the industry’s regulatory fault line

The dispute over AI is ultimately a dispute over player data. Online betting companies gather detailed information about deposits, withdrawals, bet types, frequency, losses, device use, location, response to promotions and time spent on platforms. That data can support compliance, improve user experience and identify fraud. It can also create highly individualized marketing loops.

Recent scrutiny has intensified because lawmakers and regulators are increasingly skeptical of claims that personalization is neutral. A New York Times investigation alleging that DraftKings used AI to target high-loss gamblers added political urgency to concerns that predictive tools can identify and exploit customers who are more likely to continue gambling despite losses. The allegations, and the backlash that followed, gave regulators a concrete example around which to frame rules that had previously been discussed in more abstract terms.

New York’s broader posture toward online betting also matters. The state has already been pressing against sweepstakes, prediction markets and conduct it views as unsuitable by sports betting operators. Its high-tax, high-revenue sports betting market gives the state both leverage and visibility. When New York tightens rules, national operators tend to pay attention because compliance decisions made there can influence practices elsewhere.

The same dynamic is playing out in adjacent digital markets. The Senate’s failed procedural vote on the CLARITY Act for cryptocurrency market structure underscored how contested authority remains over fast-moving digital products. State attorneys general and the Indian Gaming Association opposed that bill partly over concerns it could weaken state oversight. Though crypto and sports betting are distinct markets, both debates center on whether federal or state authorities should set the rules for technology-driven consumer risk.

Compliance stakes extend beyond marketing departments

If New York’s AI restrictions are adopted, sportsbooks would likely need to examine more than promotional copy. They would have to map where automated decision-making enters the customer journey, including onboarding, segmentation, odds displays, retention campaigns, bonus eligibility, wager suggestions and responsible gambling interventions. Vendors and affiliates would need to show how their systems use customer data and whether their models produce individualized recommendations or incentives covered by the rule.

Consent will be a central issue. The proposal would allow customers to consent to some uses, except in parimutuel horse racing. But consent in digital gambling can be difficult to operationalize. Regulators may scrutinize whether disclosures are clear, whether customers can opt out easily and whether refusing consent affects access to products or promotions. Operators that rely on broad privacy-policy language may need more explicit, product-level permissions.

The rules could also reshape competition. Larger operators may have the resources to rebuild data systems, establish AI governance committees and audit vendor tools. Smaller sportsbooks and affiliates may find the compliance burden more difficult, particularly if they depend on third-party marketing and personalization platforms. That could widen the gap between companies with mature compliance infrastructure and those built around aggressive customer acquisition.

The industry’s AI debate is entering a more consequential phase

Until recently, much of the gaming industry’s AI discussion focused on potential: faster support, better fraud detection, sharper risk scoring and more efficient marketing. The current regulatory wave is forcing a more precise question: which AI uses serve the player, and which primarily serve the operator?

That distinction will define the next phase of gambling regulation. Voluntary frameworks such as Malta’s charter may help establish governance norms, while partnerships such as Mindway AI and GamScore show how technology can be positioned around player protection. New York’s proposal goes further by challenging the use of AI in the revenue engine of sports betting: personalization designed to influence behavior.

The outcome will matter beyond one state. If New York finalizes its restrictions, other U.S. regulators could use the model to pursue similar limits, particularly in states where sports betting growth has increased concerns about advertising saturation and problem gambling. Operators may respond by adopting national standards rather than maintaining separate marketing systems by state.

For now, the industry is being pushed toward a clearer accounting of its algorithms. AI systems that identify harm, support transparency and give players more control are likely to remain defensible. Systems that predict who will spend more, lose more or respond to targeted incentives are becoming a regulatory liability.