New York Times investigation alleges DraftKings uses AI to target high-loss gamblers
A New York Times investigation claims that sports betting operator DraftKings uses artificial intelligence and machine learning technology to identify customers who are more likely to respond to promotions by continuing to gamble and lose money.
The publication reported that DraftKings began developing a machine learning model in 2023, using customer betting data to assess how likely individuals were to generate additional gambling losses following promotional offers.
The reporting says the system considered factors such as how often the user bet, their account balance and their historical losses, assigning customers scores based on how much money they were expected to generate from promotions.
The newspaper cites Jayden Butts, a former DraftKings Data Analyst who was involved in testing the model on customers, who reportedly said DraftKings wanted to use the system to redirect free bets and bonuses towards individuals who were expected to gamble and lose more.
The New York Times said the investigation was based on interviews with over 40 former DraftKings employees, alongside internal documents, Slack messages and betting data from tests conducted on users.
Former employees also allegedly told the newspaper that DraftKings had considered using similar technology to identify customers at risk of developing gambling problems, but these projects were later canceled.
DraftKings responded by saying its promotions are “directed toward customers who demonstrate sustained, engaged use of our platform, not toward customers based on their losses.”
The operator’s Chief Responsible Gaming Officer, Lori Kalani, added that the company chose not to use risk prediction technology to identify problem gamblers because there wasn’t enough evidence that it was effective.
DraftKings has said it uses AI to personalize promotional spending and that data analytics improved margins on promotion-driven sports wagers by 13% last year.
The investigation comes as DraftKings faces financial pressure, with second-quarter 2026 revenue falling 5% year-over-year to US$1.4 billion despite a 15% increase in sports betting volume to US$13.1 billion.
Dig Deeper
Related stories
2026/01/22: FanDuel announces responsible gaming campaign “Play with a Plan”
2026/09/18: US House committee advances bill to restore gambling loss deduction
2026/08/25: Total Philippines gambling ad ban seen handing market back to illegal operators, data from five countries show
2026/06/26: Into the unknown with Light & Wonder’s New Markets guru
2026/01/22: FanDuel announces responsible gaming campaign “Play with a Plan”
2026/09/18: US House committee advances bill to restore gambling loss deduction
2026/08/25: Total Philippines gambling ad ban seen handing market back to illegal operators, data from five countries show
2026/06/26: Into the unknown with Light & Wonder’s New Markets guru
The Backstory
AI has moved from trading desks to betting wallets
The allegations that DraftKings used artificial intelligence to identify customers likely to respond to promotions by gambling and losing more land at a moment when data science has become central to the economics of online betting. Sportsbooks no longer compete only on odds, market depth or celebrity advertising. They compete on customer segmentation, promotion efficiency and the ability to predict which users will keep playing after receiving a bonus.
That shift has sharpened a long-running tension in regulated gambling: the same behavioral data that can be used to manage risk, tailor promotions and improve margins also can be used to identify customers showing signs of harm. The New York Times investigation described a system built to score users based on expected promotional value, while former employees said separate efforts to use similar tools for problem-gambling detection were considered and dropped. DraftKings has disputed the characterization, saying promotions are aimed at sustained engagement rather than losses and that its responsible gaming leadership found insufficient evidence to deploy risk-prediction tools for problem gambling.
The stakes are unusually high because U.S. sports betting has matured into a scale business with heavy marketing costs, thin promotional margins and intense investor pressure. Operators are trying to reduce giveaways while keeping high-value customers active. AI offers a path to do that more efficiently, but it also raises the question regulators have not fully answered: when does personalization become exploitation?
Responsible gaming is becoming a technology contest
DraftKings’ position contrasts with the direction some rivals are emphasizing publicly. FanDuel, the U.S. market leader by many measures, recently rolled out a broader responsible gaming platform built around customer planning, account visibility and intervention tools. Its “Play with a Plan” campaign promotes dashboards, loss limits, deposit alerts and real-time customer check-ins as mainstream product features rather than compliance add-ons.
FanDuel’s approach is notable because it also relies on data and machine learning. Its Real-Time Check-In tool uses deposit information and behavioral signals to prompt customers during play. That illustrates the central policy dilemma: machine learning is not inherently protective or predatory. Its effect depends on what goal the model is optimized to achieve, who monitors it and whether customers or regulators can understand the output.
Operators increasingly describe responsible gaming tools as voluntary, empowering and friction-light. That language reflects research showing users are more likely to engage with tools framed as budgeting aids than as warnings. Yet voluntary systems also leave companies with discretion over how aggressively to intervene when a customer is highly profitable but exhibiting risky behavior. If AI can identify which gamblers are likely to respond to a free bet, it likely can identify patterns associated with chasing losses, escalating deposits or ignoring limits. The question is whether companies will be required to use those signals for protection as well as revenue.
Regulators are weighing visibility against prohibition
The broader international debate over gambling oversight shows why the DraftKings allegations matter beyond one operator. In the Philippines, lawmakers are considering whether to ban gambling advertising altogether, a proposal critics say could undermine the legal market by pushing players back to offshore sites. A detailed market analysis of the proposed ban argued that regulators made progress by channeling activity from illegal operators into licensed platforms, where age checks, transaction monitoring and exclusion tools can be enforced. The article warned that a total Philippines gambling ad ban could hand market share back to illegal operators.
That debate is relevant to AI-driven promotion because advertising and personalization are becoming intertwined. A bonus displayed inside an app, an email offer and a targeted push notification are all marketing. A broad restriction on promotion may reduce exposure to regulated brands, but it may also remove the channels through which legal operators can be monitored and sanctioned. Conversely, leaving operators free to target customers without clear rules risks creating incentives to find and retain the most vulnerable players.
The Philippine example shows the enforcement trade-off. Licensed companies can be fined, suspended or required to change conduct. Offshore operators cannot be supervised in the same way. In the United States, the risk is different but related: legal operators are visible to state regulators, but many regulatory frameworks were built around licensing, geolocation, age verification and advertising disclosures, not opaque predictive models that determine which users receive which offers.
Market pressure is pushing operators to get more precise
Sports betting’s post-legalization growth phase relied heavily on bonuses, risk-free bets and expensive customer acquisition campaigns. As markets matured, operators shifted toward profitability, cutting promotions and using analytics to spend more selectively. DraftKings’ statement that data analytics improved margins on promotion-driven sports wagers underscores how central optimization has become to the business model.
That pressure sits alongside tax and policy debates that affect customer behavior. The U.S. House Ways and Means Committee recently advanced legislation to restore the full federal deduction for gambling losses, reversing a 90% cap that industry groups warned could create taxable income even when a bettor broke even. Supporters of the bill to restore the gambling loss deduction said it would help keep players in the legal market, where operators face reporting and consumer-protection requirements.
The tax issue and the AI issue point to the same structural concern: if legal gambling becomes more punitive for customers or more distrusted because of perceived targeting, some users may migrate to unregulated alternatives. At the same time, if licensed operators are allowed to use sophisticated models without meaningful oversight, regulation may appear to legitimize practices that critics view as harmful. The industry’s argument that regulation is safer than prohibition depends on the conduct of regulated companies.
Digital gambling is blurring old product lines
The DraftKings investigation also comes as online gambling products become more immersive and closely linked to land-based experiences. In New Jersey, Awager and BetMGM launched live-streamed slot play that lets online customers control physical machines in regulated studio environments. The Awager-BetMGM live-streamed slots partnership reflects a broader push to merge the trust and familiarity of casino floors with the convenience and immediacy of mobile gambling.
Suppliers expanding into newer regulated markets are seeing that customer behavior can vary sharply by jurisdiction. Light & Wonder’s new markets strategy, including its work in the Philippines, has shown that land-based casino customers and online-first players may be distinct audiences with different habits and product expectations. The company’s emerging-markets leadership described a need to adapt quickly rather than assume that products successful in casinos will automatically resonate online, as detailed in its discussion of expansion into new iGaming markets.
That product fragmentation increases the value of customer data. Operators and suppliers want to know which game formats, bonuses and messages work for each segment. But the more granular the targeting becomes, the harder it is for regulators, researchers and consumers to see whether tools are being used to improve user experience, encourage sustainable play or intensify losses.
The next fight is over accountability, not algorithms
The central issue is unlikely to be whether gambling companies may use AI. They already do, and competitors in finance, retail, media and travel use similar systems to predict customer behavior. The more practical questions are what types of data can be used, which outcomes can be optimized, how models are audited and when companies must intervene if predictive tools identify harm.
Regulators may face pressure to require explainability for promotion models, independent testing of risk-scoring systems and clear separation between responsible gaming analytics and revenue-maximizing campaigns. Lawmakers also may examine whether customers should be told when offers are personalized based on betting history, losses, deposit patterns or account balances.
For DraftKings, the reputational risk is that the company’s most advanced technology could be portrayed as working against customers rather than protecting them. For the wider industry, the risk is larger: if operators cannot show that AI strengthens consumer safeguards as much as it improves margins, calls for stricter limits on advertising, promotions and product design are likely to grow.
