Artificial intelligence is facing increased regulatory scrutiny in the United States sports betting industry as New York and Massachusetts examine how licensed sportsbooks use customer data to personalize promotions. The developments could reshape how sportsbook software providers build marketing automation, customer analytics, and responsible gambling systems. For operators and industry observers comparing established brands and their customer-facing practices, Everything you need to know about Heritage Sports offers additional context on a long-running sportsbook in the competitive U.S. betting market.
On October 15, 2026, SBC Americas reported that Massachusetts regulators had requested detailed information from licensed operators about their use of artificial intelligence and machine learning. New York regulators, meanwhile, advanced proposals that would restrict certain forms of AI-driven promotional targeting and personalized betting recommendations.
The regulatory attention follows a September 19 investigation by The New York Times into DraftKings’ use of machine learning to assess customer responses to promotions. DraftKings disputed the characterization of its practices and defended its approach to customer engagement. The reporting nevertheless prompted broader questions about how betting platforms use predictive analytics and whether commercial optimization can conflict with player protection.
For sportsbook technology companies, the issue extends beyond marketing policy. It raises practical questions about data collection, algorithmic decision-making, audit trails, automated messaging, and the ability to prevent promotions from reaching customers who may be experiencing gambling-related harm.
Massachusetts Demands Detailed Disclosures From Sportsbook Operators
The Massachusetts Gaming Commission (MGC) has begun examining how sports wagering companies deploy AI across their customer-facing operations. According to the October 15 report from SBC Americas, Executive Director Dean Serpa sent letters dated October 9 to all licensed sportsbooks in the state, requesting written responses by November 3, 2026.
The inquiry covers several areas of sportsbook operations, including promotional campaigns, customer engagement, behavioral analytics, and the use of machine learning to predict future wagering activity.
Regulators want to know whether operators use customer losses, deposit activity, betting frequency, and other account information when deciding which promotions to distribute. They are also examining whether automated systems can initiate marketing communications without direct human approval.
These questions matter because modern sportsbook platforms can process large volumes of behavioral data and use predictive models to estimate how customers might respond to different offers. A system can rank users by predicted engagement, calculate the likely commercial return of a promotion, and distribute targeted incentives through email, push notifications, or in-app messaging.
The same capabilities can support responsible gambling programs, but their effects depend on the objectives and safeguards built into the software.
MGC Chairman Jordan Maynard had signaled the commission’s interest in the subject during a September 24 meeting. The commission also indicated that it planned to interview representatives of individual operators to inform its assessment of AI use in Massachusetts.
For software vendors supplying licensed operators, the inquiry creates a clear operational priority: demonstrate how automated marketing decisions are made, which data influences them, and what controls prevent potentially harmful targeting.
New York Advances Proposed Restrictions on AI-Driven Promotions
New York is taking a more prescriptive approach. On October 16, 2026, the New York State Gaming Commission advanced revised proposals addressing AI-assisted promotional targeting and personalized wagering recommendations. The developments were reported by SBC Americas on October 15 and by PlayNY on October 18.
The proposed measures would restrict operators from using customer data and algorithmic profiling to personalize certain promotions or recommend wagers to individual customers. They would also prohibit specified promotional practices intended to increase customer spending, revenue, or engagement, subject to the details of the proposed rules.
The proposals are not a final, fully implemented ban. They must undergo another public-comment process before the commission can consider possible final adoption. Operators and technology providers therefore need to distinguish current legal obligations from proposed requirements that may change during the rulemaking process.
New York’s approach also includes a separate player-protection proposal. The commission is seeking measures that would require operators to identify customers displaying signs of potentially harmful gambling behavior and designate personnel to connect those customers with responsible gambling tools and resources.
The implications for sportsbook platforms could be significant. Systems that currently use a customer’s betting history to recommend markets or trigger incentives may require additional eligibility checks, configurable restrictions, and documented consent processes.
A sportsbook operating in multiple states could also face different requirements for the same marketing feature. Its platform would need to apply the appropriate rules based on a customer’s jurisdiction, account status, and applicable permissions.
Why DraftKings Put Sportsbook AI Under the Microscope
The regulatory response follows an investigation published by The New York Times on September 19, 2026, examining DraftKings’ use of machine learning in promotional decisions. The newspaper reported that the company developed a model in 2023 to estimate which customers were more likely to respond to incentives by gambling and losing more.
According to the investigation, the model analyzed customer activity, including betting frequency, account balances, and historical losses. Former employees interviewed by the newspaper raised concerns that promotional targeting could favor customers whose behavior made them more commercially valuable, even when similar behavioral information might help identify signs of gambling-related harm.
The newspaper also reported that some former employees believed efforts to develop predictive tools for identifying problem gambling had been sidelined. DraftKings disputed the suggestion that its business objectives conflicted with responsible gambling and defended its customer-protection practices.
The distinction between the reported allegations and the company’s position is important. Predictive analytics are not inherently harmful, and the existence of an AI-driven promotional system does not by itself establish misconduct. The central question is how the system is trained, what outcomes it optimizes, and which protections constrain its decisions.
For example, a model trained to maximize the probability that a customer accepts a promotional offer may produce different results from one optimized to improve long-term engagement while excluding customers who display high-risk behavioral indicators.
Both systems may use similar inputs. Their objectives, constraints, and consequences can differ considerably.
This is where software governance becomes essential. Operators need to understand which variables drive model outputs, whether those outputs can be explained, and whether responsible gambling controls can override a marketing recommendation.
How AI Marketing Works Inside a Sportsbook Platform
AI-driven marketing typically combines customer data, predictive models, segmentation tools, and automated communication systems. A sportsbook might use these components to determine which users receive an offer, when an offer appears, and which message is displayed.
A typical workflow can involve several connected processes:
- Data collection: The platform processes permitted account activity, transaction records, betting patterns, and customer preferences.
- Predictive scoring: A model estimates likely responses to promotional messages or predicts future account activity.
- Eligibility checks: The system evaluates whether the customer can receive a particular offer under applicable rules and responsible gambling restrictions.
- Campaign execution: An approved offer is delivered through an authorized channel, with the decision recorded for later review.
The most important distinction is between predicting customer behavior and deciding what the operator should do with that prediction.
A system that identifies a change in betting frequency could trigger a review or offer a responsible gambling resource. A commercially focused model could instead interpret the same signal as an opportunity to encourage further betting. The underlying analytics may be similar, but the intended response is different.
Software vendors should therefore avoid treating responsible gambling as a separate feature that operates independently of marketing. Eligibility checks, exclusion lists, consent settings, and risk indicators should be integrated into the decision process before a promotional message is released.
For businesses evaluating sportsbook software solutions, the ability to configure these controls across marketing, analytics, and account management is becoming an important consideration in platform selection.
Five Technical Priorities for Sportsbook Software Providers
The developments in New York and Massachusetts provide a useful framework for evaluating the technical readiness of sportsbook platforms. Although the precise legal requirements remain subject to the rulemaking process, several capabilities can help operators prepare for closer oversight.
1. Configurable State-Level Compliance Rules
Sportsbook platforms should support jurisdiction-specific controls rather than rely on one global marketing configuration.
A compliance engine could determine whether an offer is permitted based on the customer’s verified location, the applicable state rules, account restrictions, and the type of promotion. If a state prohibits a particular form of AI-assisted targeting, the platform should be able to disable that feature for the relevant users without unnecessarily disrupting permitted services elsewhere.
Rules should be version-controlled so operators can identify which configuration applied when a particular message was generated.
2. Explainable Promotional Decisions
Operators need records showing why a customer received a promotion and which system components influenced that decision.
An auditable record could include the model version, relevant eligibility checks, campaign rules, approval status, and timestamp. Where appropriate, it should also identify whether a human reviewed the decision or whether the system executed it automatically.
This information can help compliance teams investigate complaints, respond to regulators, and identify unintended behavior.
3. Responsible Gambling Controls Before Delivery
Risk controls should be applied before an offer reaches a customer, not only after a campaign has been sent.
If an account is subject to self-exclusion, a marketing restriction, or another applicable protective measure, the promotional system should block ineligible communications. A customer who triggers a risk indicator should not automatically become eligible for a retention campaign simply because the predictive model forecasts a high response rate.
These controls should be tested across push notifications, email, SMS, in-app messages, and third-party campaign tools to prevent inconsistent enforcement.
4. Human Oversight and Model Governance
Automated marketing should have clear accountability. Operators should document who approves a model, who can change its objectives, and who is responsible for investigating unexpected results.
Software teams should also test models for unintended incentives. For instance, a model designed to improve promotional efficiency might disproportionately select customers with frequent losses or escalating deposits if those variables are strongly associated with campaign response.
Monitoring should track not only conversion rates and revenue but also exclusion failures, complaints, risk indicators, and the distribution of offers among customer groups.
5. Reliable Data Lineage and Vendor Documentation
Many operators rely on external vendors for customer relationship management, analytics, messaging, and personalization. That makes it necessary to understand which systems receive customer data and how they use it.
Providers should maintain documentation identifying the source of each relevant data field, the model or rule that consumes it, and the downstream actions it can trigger. Contracts and technical interfaces should also clarify responsibilities for data access, retention, security, audits, and regulatory requests.
Without that visibility, an operator may struggle to explain a decision made by a third-party model even though the operator remains responsible for its licensed activity.
The Business Impact: Marketing Flexibility Versus Regulatory Risk
AI can help sportsbook operators improve campaign efficiency, organize customer support, and allocate marketing budgets. Restrictions on certain targeting practices could reduce some of that flexibility, especially for platforms built around individualized promotions.
However, regulatory uncertainty also creates an opportunity for software providers that can demonstrate reliable compliance controls. Operators may favor systems that let them adjust promotional rules quickly, monitor automated decisions, and document the reasons behind each action.
The commercial challenge is to distinguish useful personalization from practices that create unacceptable player-protection risks.
For example, a sportsbook could use machine learning to recommend relevant educational content, improve customer-service routing, or detect suspicious account access. Those applications differ from a model that uses a customer’s losses to determine whether that person should receive an incentive to wager again.
The distinction will depend on the particular system, its purpose, the data involved, and the applicable rules. AI use should not be evaluated as a single category in which every application receives the same treatment.
Operators should also avoid assuming that consent alone resolves every compliance issue. Even where a proposed rule permits customer consent for certain activities, separate restrictions may still apply to promotions, self-exclusion, and responsible gambling obligations. Legal teams will need to assess the final wording of each state’s rules before approving a deployment.
What Sportsbook Operators Should Prepare Before November 3
The Massachusetts deadline gives operators a concrete near-term task. Licensed sportsbooks should prepare accurate disclosures of their AI and machine-learning systems before November 3, 2026, covering the questions raised by the Massachusetts Gaming Commission.
A practical readiness review should include a complete inventory of marketing algorithms, data sources, automated communication triggers, and third-party tools. Compliance teams should work with engineering, data science, marketing, and responsible gambling personnel to establish how each system functions.
Operators should also identify whether promotional decisions can be made without human review, which risk indicators are considered, and what safeguards prevent vulnerable customers from receiving inappropriate offers.
For New York, technology teams should track the next public-comment stage and avoid hard-coding assumptions about proposals that have not yet become final rules. A configurable architecture will make it easier to respond if the commission adopts restrictions or modifies the requirements.
The same preparation can benefit operators in other states. Massachusetts and New York are among the jurisdictions examining AI-driven marketing, and SBC Americas has also reported interest in the issue from Maine, Maryland, and federal lawmakers. The exact policies differ, so a single national compliance assumption would be premature.
Sportsbook AI Regulation Is Becoming a Software Design Issue
The scrutiny in New York and Massachusetts illustrates how regulatory requirements can affect the architecture of sportsbook technology. Marketing automation, customer analytics, and responsible gambling systems can no longer be treated as unrelated components when the same behavioral data influences all three.
The immediate milestones are clear: Massachusetts sportsbooks must respond to the commission’s inquiry by November 3, 2026, while New York’s proposed restrictions continue through the rulemaking process. The eventual requirements will determine how much operators must change their systems, but the direction of scrutiny is already evident.
For sportsbook software providers, the most useful response is to build systems that support transparent decision-making, jurisdiction-specific controls, effective risk exclusions, and verifiable audit records. These capabilities can help operators adapt to changing requirements without rebuilding their entire marketing stack each time a regulator changes the rules.
The long-term test for AI in sports betting will not be how accurately a platform predicts customer behavior alone. It will also be whether the platform can demonstrate that its predictions are used within clear operational, legal, and player-protection boundaries.