DraftKings Uses AI to Target Gamblers Likeliest to Lose, Former Employees Reveal

DraftKings Builds Machine Learning Models to Identify High-Loss Gamblers

DraftKings developed machine learning models in 2023 to optimize its hundreds of millions of dollars in annual spending on promotional incentives. The online gambling platform tasked data analyst Jayden Butts with testing a model designed to answer a specific question: which customers would respond to promotions by gambling and losing more money1

1

. The AI system scored each customer based on betting records, with higher scores indicating gamblers likely to lose more money for each promotion offered2

2

.

The model analyzed dozens of data points for each gambler, including betting frequency, daily account balances, typical loss amounts compared to bet sizes, and likelihood of stopping gambling1

1

. Six former employees who worked on these AI-driven strategies confirmed that DraftKings has continued refining its methods to target losing gamblers with promotional incentives that encourage more betting2

2

.

Ethical Concerns Emerge Among Former Data Analysts

Butts quickly recognized troubling implications in the assignment. "We are looking for traits and features that we can target that indicate a good investment," he said. "By strict financial logic, the best investment would be a problem gambler"1

1

. His concern centered on a fundamental business reality: DraftKings makes money when gamblers lose money, and the model specifically identified those it could get to lose the most2

2

.

A New York Times investigation interviewed more than 40 former DraftKings employees and obtained internal documents including research memos, presentations, Slack messages, and betting records from customer experiments1

1

. Butts and five other former employees who worked on promotional targeting told The Times they regretted building technology they now viewed as dangerous2

2

.

Risk Prediction Technology Sidelined While Loss-Maximizing AI Advances

While DraftKings refined its AI to target gamblers most likely to lose, four former employees said the company stalled or squashed efforts to use similar technology for responsible gaming purposes1

1

. Employees developed a machine learning model that would have assigned users risk scores to identify potential problem gambling based on betting activity, but the company sidelined it according to two former employees who worked on that project2

2

.

The same betting records that power promotional targeting may contain signs that a person is headed for gambling addiction risks1

1

. This creates a stark contrast: the online gambling platform actively uses machine learning models to identify engaged users who will lose more money while declining to implement technology that could flag those developing harmful patterns.

Company Response and Continued AI Development

DraftKings fired Butts in late 2024 for performance reasons amid a business downturn that led some to question whether promotional experiments were working as intended1

1

. The company briefly paused some data science work before resuming with new machine learning projects2

2

.

In a statement, DraftKings rejected implications that its marketing practices are unfair or improperly target customers, stating promotions are "directed toward customers who demonstrate sustained, engaged use of our platform, not toward customers based on their losses"1

1

. Lori Kalani, DraftKings's chief responsible gaming officer, said the business depends on "customers who are betting within their means, are betting for entertainment and betting for fun"2

2

.

Kalani acknowledged DraftKings monitors customers for potentially risky behaviors but confirmed the company declined to use risk prediction technology, saying there wasn't evidence it was helpful1

1

. A former analyst who quit in 2024 described the predatory nature of the system: "If you lose more, we give you more, so you keep playing more"2

2

.