The best investment would be a problem gambler. That line did not come from a protest sign. It came from a former DraftKings data analyst describing the logic of a model his own company built.
Jayden Butts told The New York Times, in the investigation published September 19, 2026: “We are looking for traits and features that we can target that indicate a good investment.” By strict financial logic, he added, “the best investment would be a problem gambler.” An unnamed former analyst put it flatter: “It is as predatory as it sounds.” I am not inventing the charge. I am reading the internal record the Times assembled from research memos, Slack messages, customer betting data, and interviews with more than 40 former employees. DraftKings disputes the framing and says it does not target on losses or problem-gaming markers. The Massachusetts Gaming Commission has not found wrongdoing. The documents still sit on the table.
In 2023 the company stood up a machine-learning model that scored customers on an internal “elasticity” metric. Higher scores meant a player was more likely to answer free bets and bonuses by wagering more and losing more. Frequency, daily balance patterns, loss-to-wager ratios. Former analyst Jayden Butts tested it first on roughly 5,000 casino players in September 2023 and pushed promotions toward the higher-scoring accounts. Six former employees told the Times the company kept refining that approach. Four others said parallel work to flag customers at risk of problem gambling, including efforts by data scientist Nestor Hernandez around 2024, got stalled or shelved. DraftKings leaned instead on an external tool, Mindway AI’s Gamalyze, for some responsible-gaming functions. The promotion engine stayed hot. The intervention engine did not.
Massachusetts Already Wrote the Rule DraftKings Is Now Being Asked to Explain
On September 24, during a public meeting of the Massachusetts Gaming Commission, Chair Jordan Maynard ordered staff to engage DraftKings over the Times reporting and to map how every licensed operator in the state uses artificial intelligence for promotions, acquisition, and responsible gaming. “These A.I. technologies are evolving rapidly across our society, and we share the concern over their application,” Maynard said. He instructed Executive Director Dean Serpa to “engage with the operator to understand the specifics that were reported.” Commissioner Paul Brodeur called the findings “troubling” and said the commission needed to “find out what the facts are on the ground.”
This is not a blank-slate jurisdiction. Massachusetts already prohibits operators from using AI or similar automated systems known or reasonably expected to make a gaming platform more addictive. Operators must also report on customer-behavior analysis tied to responsible gambling. The commission had already commissioned a UNLV study, released around November 2025, that flagged a governance gap and spun up an internal AI task force. DraftKings is headquartered in Boston and holds a sports betting license there. The probe lands in the company’s own backyard, under rules the company already lives under. No enforcement action on AI had been taken before Maynard spoke. That is the point. The statute existed. The model still ran.
The Margin Math Explains Why Nobody Hit the Brakes
DraftKings spends hundreds of millions of dollars every year on promotional incentives. Company disclosures and investor materials cited around the Times reporting put roughly $400 million in AI-automated and personalized promotions into the market in 2025. AI targeting was credited with lifting promotion-driven sportsbook margins by 13 percent that year. Gross gambling revenue, sports and casino, sat near $8.7 billion in 2025 against roughly $3 billion in promotions, per research the Times cited. Those are not abstract ledger lines. They are the reason an elasticity score that treats heavy losers as high-ROI targets looks, inside the building, like good business.
I bet. Plenty of people reading this bet. The free bet that lands in your account after a rough weekend is not a gift from a friend. It is a calculated instrument. When the same shop that can score you as “likely to lose more” also slows-walks the tools meant to catch the people sliding into trouble, the relationship stops being entertainment and starts looking like extraction. National Council on Problem Gambling figures put more than 31,000 Americans on its helpline every month. That volume does not appear out of thin air. It appears when incentives are optimized for volume of play and the safety net is treated as optional infrastructure.
DraftKings will keep saying the analysis was preliminary and that it does not target problem gamblers. The commission will gather facts. Other operators in Massachusetts, FanDuel included, will get the same questions. I do not need a verdict sheet to see the design choice the Times laid out: score the people most responsive to losing more money, flood them with bonuses, and leave the predictive responsible-gaming work on the shelf. That design choice is now on the public record in the state where the company is based.
The elasticity model was built to find a return. Massachusetts is finally asking what that return cost the people who funded it.
21+. Commentary, not betting advice. Gambling problem? Call or text 1-800-MY-RESET (in Wisconsin, 1-800-GAMBLE-5).
