DraftKings AI Model Targets Chronic Losing Gamblers With Promotional Ads

What happened

Why it matters

• ·Behavioral data collected for one lawful purpose (account operation, fraud prevention) can be repurposed to train models that target vulnerable customers. Existing data minimization and purpose-limitation controls under frameworks like the Final CPPA Regulations on Cybersecurity, Risk Assessments, and Automated Decision-Making Technologies may not block this if the data use is deemed consistent with the original collection context.
• ·This case signals growing regulatory interest in AI-powered personalized marketing that targets people showing signs of harm. Compliance teams in gaming, financial services, insurance, and health tech should assess whether their model-driven retention or re-engagement programs could be characterized as predatory targeting under consumer protection or unfair-practices standards, including FTC Enforcement on AI (Section 5 of the FTC Act).
• ·The EFF analysis underscores that AI amplifies the harm of behavioral advertising by enabling precision targeting at scale with no incremental data cost. This exposes a gap in existing fairness and bias monitoring controls. Those controls check for outcome disparity across demographic groups. They do not check for systematic targeting of individual vulnerability states identified through predictive modeling.

Governance controls affected

What to do now

• ☐Map every machine learning model used in customer marketing, retention, or re-engagement to the underlying data inputs: flag any model trained on loss history, financial distress signals, or behavioral patterns that could classify customers by vulnerability.
• ☐Review your stated purpose for collecting customer behavioral data against the downstream uses those data inputs currently feed, and identify any model application where the use would surprise a reasonable customer who read your privacy notice at sign-up.
• ☐Ask your data science and marketing teams whether any AI-driven campaign segmentation or personalized outreach specifically targets customers who have stopped engaging or who show high-loss patterns, and document whether human review is required before those segments receive promotional content.
• ☐Assess whether your AI fairness and bias monitoring program checks for disproportionate targeting of vulnerable populations (not just protected class disparities), and update your evaluation criteria if that check is absent.
• ☐Engage legal counsel to evaluate your customer-facing AI marketing uses against FTC unfair-practices authority and any applicable state automated decision-making or consumer protection rules, particularly if your business operates in gaming, lending, insurance, or health-adjacent categories.

What to watch next

Regulators are increasingly treating AI-powered behavioral targeting of vulnerable populations as a consumer harm issue, not merely a data privacy one. The FTC Enforcement on AI (Section 5 of the FTC Act) provides existing authority to act on unfair or deceptive practices without waiting for new AI-specific legislation. State-level automated decision-making rules are advancing in California and Colorado. They are beginning to require impact assessments for AI systems that affect consumer access to products and services. This could extend to retention and re-engagement marketing. Compliance teams should monitor whether the DraftKings analysis draws a formal regulatory response and whether problem-gambling advocacy groups file complaints that prompt FTC or state attorney general review.