NYT: DraftKings Built AI to Target Its Likely Biggest Losers
A New York Times investigation published this week reported that DraftKings built a machine-learning system in 2023 to identify which customers were likely to lose the most money after receiving promotional offers — and used it to help direct roughly $400 million in automated promotions in 2025 alone. The report, based on interviews with more than 40 former DraftKings employees and internal company documents, has drawn fresh attention to how sportsbooks target bonus bets and free-play offers.
What the model reportedly did
According to the Times' reporting, the internal system scored users on "elasticity" — how responsive a given customer was likely to be to free bets and other promotions, measured against behavioral signals including playing frequency, account balance patterns, and loss-to-wager ratios. Former data analyst Jayden Butts was quoted describing the underlying logic bluntly: identifying users whose promotional response would generate more revenue for the company than the cost of the promotion itself, and then, in his words, being able to "open the floodgates" with offers toward those users. He separately told the paper the approach is "as predatory as it sounds."
The report notes that the same behavioral signals used to build the elasticity model — playing frequency, balance swings, and loss ratios — overlap with indicators clinicians use to identify problem gambling.
Harm-detection efforts reportedly shelved
The investigation also describes a separate, unrelated effort: a data scientist built a model intended to flag customers showing signs of a gambling crisis, using much of the same underlying data infrastructure as the elasticity model. According to the reporting, that project and at least one other similar harm-detection effort were shelved rather than deployed, with the company's Chief Responsible Gaming Officer, Lori Kalani, telling the Times the risk-modeling approach wasn't sufficiently "evidence-based" to act on.
DraftKings' response
DraftKings disputed the characterization in the Times' report, telling the paper its promotions are aimed at customers showing "sustained, engaged use" of the platform rather than customers selected because they're likely to lose money. The company pointed to its existing responsible-gaming tools, including deposit limits and self-exclusion options, as evidence of its broader approach to player protection.
Why it matters to bettors and regulators
The report lands amid growing legislative attention to how sportsbooks use data and automation in marketing. The federal SAFE Bet Act, currently before Congress, would restrict AI-driven personalization of gambling promotions, and several states have introduced comparable proposals. Coverage of the investigation also noted DraftKings increased its federal lobbying spending to roughly $900,000 in 2025. For bettors, the report is a reminder that "personalized" bonus offers are generated by systems optimizing for the operator's return, not the player's — worth keeping in mind alongside the deposit limits and self-exclusion tools every legal operator is required to offer. See our responsible gambling resources for how to set those controls on your own account.