DraftKings: Leveraging AI to Predate on Chronic Gamblers

Originally published by the Electronic Frontier Foundation (EFF) via the Deeplinks Blog Author: Devanshi Nishar | Date: September 24, 2026

DraftKings, a titan in the online sports betting industry, is currently employing artificial intelligence to identify and target the specific subset of its user base most likely to lose money and succumb to gambling promotions. This strategy is a sophisticated application of online behavioral advertising—the practice of tailoring advertisements based on a comprehensive profile of data harvested from the user.

The Mechanics of Predatory Targeting

The core of DraftKings' strategy involves a feedback loop where user data is weaponized to maximize corporate revenue. According to reports from the New York Times, the company utilizes historical betting records to train a machine learning model.

The logic is simple: DraftKings has a massive financial incentive to ensure that "losing gamblers" remain active, as these individuals are the primary drivers of the company's profit margins.

The Human Cost: Vulnerability as a Product

The most alarming aspect of this AI-driven approach is its impact on "problem gamblers." These are individuals who continue to gamble despite severe negative consequences regarding their:

• Financial stability
• Personal relationships
• Mental and physical health

Instead of utilizing their data to trigger interventions or mitigate risk, DraftKings uses it to capitalize on these vulnerabilities.

Why AI Supercharges the Harm

While behavioral advertising has always been problematic, the integration of AI has magnified the danger. AI operates as a black box, meaning the developers often cannot pinpoint exactly which data points are driving the model's decisions. This creates a perverse incentive to collect as much data as possible to refine the algorithm.

The "Profit Equation" for such a model can be conceptualized as: $\text{Corporate Profit} \approx \sum (\text{Vulnerable Users} \times \text{AI-Driven Retention Rate} \times \text{Loss Probability})$

The Surveillance Pipeline

The data harvested for these ads does not stay within the betting app. The surveillance industry thrives on this information, which is frequently sold to:

1. Financial Institutions (Banks)
2. Insurance Providers
3. Government Agencies (e.g., Customs and Border Protection - CBP)

"ICE is also taking an interest in the data fueling ad tech: earlier this year, ICE published a Request for Information 'seeking information to better understand how the industry’s commercial Big Data and Ad Tech providers can directly support investigations activities.'"

The Policy Gap: First-Party vs. Third-Party Data

A critical nuance in the DraftKings case is that they appear to rely exclusively on "first-party data" (information collected directly from their own users). This renders many current legislative efforts ineffective.

Limit the sale of third-party data to brokers $\rightarrow$ This is not enough.

If a company can target you using only the data they collected from you, the harm persists regardless of whether they sell that data to a third party.

The EFF's Stance and Solution

The EFF argues that the only way to stop this cycle is a total ban on all behavioral advertising. By removing the ability to send personalized ads, the incentive for companies to engage in mass behavioral surveillance disappears.

How to Protect Yourself

If you want to reclaim control over your digital footprint, consider the following steps:

• Explore the Surveillance Self-Defense project.
• Review privacy settings on all mobile applications.
• Implement tools to block trackers on websites.
• Use the Privacy Badger extension.

# Conceptual logic of a predatory ad trigger
def target_user(user_profile):
    if user_profile.is_losing_streak == True and user_profile.responsiveness == 'High':
        send_promotion(user_profile.id, "Come back and win big!")
    else:
        pass # User is too profitable as a winner or not susceptible

---

Visual References from the Original Report: