DraftKings trains AI on betting records to target losing gamblers, EFF says

What the reporting says

According to the Electronic Frontier Foundation (EFF), online sports betting operator DraftKings is using artificial intelligence to find the customers most likely to place losing bets and to respond to gambling promotions. Writing on the EFF's Deeplinks blog, the digital rights group cites New York Times reporting that DraftKings trains a machine learning model on its customers' own betting records to identify gamblers who tend to lose, then sends those users targeted advertising and promotions designed to pull them back onto the platform to keep betting.

The EFF lays out the underlying commercial logic: losing gamblers are the customers who actually generate DraftKings' revenue, so the company has a structural incentive to keep them active. The practical consequence, the group argues, is that people who would be classified as problem gamblers — those who keep gambling despite harm to their finances, relationships and wellbeing — are among the most likely to be flagged by the model and re-engaged with tailored offers. Rather than reducing risk for vulnerable users, the system profits from their vulnerability.

How AI scales behavioral advertising

For the EFF, DraftKings is one instance of a wider problem with online behavioral advertising, the practice of personalizing ads using data collected about individual users. Adding AI to that pipeline, the group argues, magnifies the harm in several ways: it incentivizes collecting even more data to train and refine models; the black-box nature of machine learning means the people building the systems often cannot predict which data points will prove useful, which pushes them to keep gathering everything; and AI lets companies process enormous datasets far faster than was previously possible.

The downstream effects extend well beyond one betting app. Data gathered for ad targeting, the EFF notes, feeds the wider surveillance economy, with such data sold to insurance companies, banks and law enforcement agencies including Customs and Border Protection. The group also points to a Request for Information that Immigration and Customs Enforcement published earlier this year asking how commercial big data and ad tech providers could directly support investigations.

The first-party data loophole

One detail in the DraftKings case carries particular weight for policy. According to the EFF, the company appears to rely solely on first-party data — information collected directly from its own users — rather than buying additional data from third parties to fuel its model. The EFF uses this to argue that regulation focused solely on limiting third-party data sharing and selling would not have prevented this kind of predatory targeting. Its conclusion is that online behavioral advertising should be banned outright: if companies cannot send personalized ads, they lose much of the incentive to collect the behavioral data that powers them.

The EFF also points readers toward its own resources, including the Surveillance Self Defense project, for people who want to limit how much data they expose through mobile apps and websites.

Why it matters

The story shows AI applied to a category where the conflict of interest is unusually explicit: a betting platform has a direct financial stake in its customers losing money, and a model trained to identify likely losers turns that stake into an automated re-engagement machine. It also highlights a live regulatory gap. As policymakers concentrate on third-party trackers and data brokers, first-party data gathered inside a company's own product remains largely unregulated — and machine learning makes that data more actionable than it has ever been. The DraftKings example concerns gambling, but the same pattern of behavioral profiling and targeted re-engagement runs through much of the modern app economy, which is why the EFF treats it as evidence for a broad ban rather than a narrow fix.