NYT: DraftKings used a machine-learning 'elasticity' score to aim promotions at bettors most likely to keep losing; Massachusetts opens review
A New York Times investigation published Sept 19, 2026 reported that DraftKings built a 2023 machine-learning model scoring online-casino customers by the losses a promotional dollar would generate (an "elasticity" score) and used AI to personalize about $400M of promotions in 2025. Former staff said the score in effect picked out problem gamblers. DraftKings denied targeting anyone based on losses. The Massachusetts Gaming Commission said on Sept 24 it would examine sportsbooks' AI use, and the EFF called for a ban on behavioral advertising.
Key facts
- NYT sources: internal memos, presentations, betting records and interviews with more than 40 former employees (as summarized by AI Weekly/Tech Times)
- Model inputs: how often a customer played, balance movements, amounts bet and lost, likelihood of quitting
- About $400M of 2025 promotional spending personalized with AI; data science credited with a 13% margin lift on promotion-driven sportsbook bets (secondary summaries of the NYT)
- Former analyst Jayden Butts: 'The best investment would be a problem gambler'
- A 2024 harm-detection model to flag at-risk users was shelved
- DraftKings: it 'does not use A.I. to target anyone based on losses and does not market to customers based on indicators of potential problem gaming'
- MGC chair Jordan Maynard: staff will engage with DraftKings to 'understand the specifics' of its AI and machine-learning practices
- EFF (Sept 24): 'all behavioral advertising should be banned'
What happened
The NYT investigation itself was not fetchable (paywall), so the figures above come from secondary summaries and are marked confidence medium. The regulator's statement and DraftKings' denial are quoted from Yogonet.
Why it matters
A concrete case of ordinary machine learning optimized for revenue learning to target vulnerable people. It feeds the push to regulate AI-driven personalization, in gambling (the SAFE Bet Act) and in advertising more broadly, just as chatbots add ads.
Changelog
- 2026-09-30: created (sweep 2026-09-29: Techmeme Sept 24 + HN 519 points)
Sources (5)
- pressNYT: Massachusetts regulator to examine DraftKings' use of AI (Sept 24)
- pressYogonet: Massachusetts Gaming Commission to examine sportsbooks over use of AI in betting promotions
- pressTech Times: DraftKings' promotional AI scored bettors by addiction markers, NYT finds
- discussionEFF: DraftKings is using AI to supercharge the harms of online behavioral advertising
- discussionHacker News discussion (519 points)
id: 2026-09-19-nyt-draftkings-ai-targets-losing-gamblers · updated 2026-09-30 · open in the interactive timeline