Losing gamblers pushed to bet more by DraftKings’ AI, report says

Executive Summary

A machine learning model used by DraftKings reportedly targeted users who were likely to gamble and incur losses through the delivery of betting promotions. The model analyzed customer data, including betting frequency, losses, and daily account balances, to predict who would respond to these advertisements. An experiment conducted by a journalist involved an individual who reported receiving more promotions as their losses increased. DraftKings denied unfair targeting, asserting that its promotions reward engagement rather than losses. The system was developed starting in 2023 and replaced a basic behavioral tracking system, utilizing an "elasticity" score to identify customers likely to lose more after receiving offers. Attempts by other employees to develop a similar model aimed at flagging compulsive gambling were halted internally. Digital rights advocates suggest that this AI capability supercharges the harms of online behavioral advertising, particularly when it encourages addictive behaviors like gambling.

Facts Only

* A machine learning model targeted people likely to gamble and lose more money through betting promotions.
* The model crunched data on customer betting frequency, losses, and daily account balances.
* DraftKings used promotions, including profit boosts on winnings, to entice users into betting more.
* An experiment involved a journalist who reported receiving more promotions as losses mounted.
* After losing nearly $1,800 on basketball bets in one evening, the journalist was invited to a VIP tryout and later full membership.
* DraftKings denied unfair targeting, arguing promotions reward engagement over losses.
* The machine learning model was developed starting in 2023.
* The system created an “elasticity” score to identify customers likely to lose more in response to offers.
* Other employees attempted to develop a model to flag users at risk of compulsive gambling, which was shut down after a demo.
* The World Health Organization estimates 1.2% of the world’s adult population has a gambling disorder.

Full Take

The deployment of predictive algorithms in marketing shifts the dynamic from simple observation to calculated inducement. The core mechanism involves using private behavioral data not just to track history but to engineer future loss by calculating individual susceptibility to promotional incentives, which reveals a structure optimized for maximizing engagement and revenue derived from risk. The failure to pursue an internal model aimed at identifying high-risk users for intervention suggests a structural conflict between commercial optimization goals and social responsibility objectives within the organization itself. This setup creates a powerful feedback loop where systems designed for profit are inherently positioned to exploit vulnerabilities, which digital rights advocates correctly identify as supercharging behavioral harm. The tension lies in the gap between the company's stated defense—that promotions reward engagement—and the measurable outcome of using the AI specifically to maximize losses across a vulnerable population. How can regulatory and ethical frameworks account for predictive systems designed explicitly to amplify known psychological vulnerabilities rather than mitigate them? What is the responsibility when a mechanism, even one created by insiders, is demonstrably deployed to push users toward states of increased financial distress?

From the original · Malwarebytes Labs

Two separate investigations have raised concerns about betting site DraftKings’ marketing to target problem gamblers, including its use of AI. On September 19, the New York Times reported that the company’s machine learning model targeted people who were more likely to respond to betting promotions by gambling and losing more money.

Read the full story at malwarebytes.com

Sentinel — Human

Confidence

The text reads like a well-researched journalistic synthesis connecting a specific AI application in gambling marketing to broader societal concerns about addiction, exhibiting the texture of human investigative reporting.

Signals Detected

Sentence length variance shows some natural variation; vocabulary is journalistic but flows reasonably well.

Maintains a consistent focus on the DraftKings case while smoothly pivoting to broader societal warnings regarding gambling and AI advertising.

Uses established journalistic structuring (setting up conflict, presenting evidence, providing context) typical of investigative reporting; no obvious verbatim repetition.

Citations to specific reports (NYT, ProPublica, EFF, WHO) and named individuals/entities suggest grounding in real-world data, even if the presentation is synthesized.

Human Indicators

The inclusion of specific, nuanced details regarding internal corporate maneuvers (e.g., shutting down the risk-flagging model) and named sources (Jake Pearson, Jayden Butts) suggests an investigative foundation.

The shift from a specific corporate case to a broad public health warning feels like editorial framing rather than pure data dumping.