DraftKings Uses AI to Target Problem Gamblers: The Ethical Crisis of Behavioral Advertising

When AI Becomes a Gambling Company's 'Precision Harvesting' Tool

A recent in-depth investigation published by the Electronic Frontier Foundation (EFF) has uncovered a troubling practice: online gambling giant DraftKings is using artificial intelligence to serve behaviorally targeted ads to heavy gamblers who show signs of addiction. The report sparked heated debate on Hacker News, earning 276 upvotes and nearly 200 comments — a clear signal of the tech community's widespread concern about AI being weaponized for "precision harm."

Behavioral advertising itself is nothing new. Internet companies routinely track browsing habits, click behavior, and purchase history to serve "relevant" ads. But when this machinery is applied to the gambling industry and supercharged with AI's predictive capabilities, the potential for harm escalates dramatically. The EFF's core argument is that AI doesn't just make ads more "targeted" — it allows gambling companies to systematically identify and exploit their most vulnerable users.

How AI 'Reads' a Gambler's Weaknesses

Traditional gambling marketing relied on broad-brush ad bombardment. AI-driven behavioral targeting is an entirely different beast. By analyzing data signals such as betting frequency, changes in wager size, login timing, and chasing losses behavior, machine learning models can construct highly granular user profiles — identifying who is already in, or sliding toward, pathological gambling.

The crux of the problem is a fundamental misalignment of incentives. For gambling platforms, heavy addicts are precisely the most profitable users — they account for the vast majority of platform revenue. As a result, AI systems are optimized not to "help users gamble responsibly," but to "maximize user engagement and total bets placed." This means the system actively identifies users showing signs of addiction and targets them with free bets, cashback offers, and "one more round" nudges.

This mechanism creates a brutal feedback loop: the more out of control a user becomes, the more the system flags them as a "high-value target," and the more intensely they are bombarded with inducements. AI here is not a neutral tool — it is an engine driven by commercial objectives, engineered to exploit human psychological vulnerabilities.

Chasing losses is one of the hallmark signals of pathological gambling — the psychology-driven behavior where a gambler, rather than cutting their losses, increases their bets after a loss in hopes of "winning it back." This phenomenon is closely tied to loss aversion in behavioral economics: humans feel the pain of a loss far more acutely than the pleasure of an equivalent gain, causing emotional responses to override rational judgment. For AI systems, chasing behavior is relatively easy to detect from data: a sharp spike in bet amounts over a short period, a shift in odds preference from conservative to aggressive, late-night logins or unusually long continuous sessions — these patterns form feature vectors that machine learning models can readily capture. The troubling reality is that this same data could be used to trigger "cooling-off" reminders or account spending limits. Instead, under a profit-maximization objective function, it is used to push "deposit bonuses" and limited-time offers — converting users at the exact moment they are most emotionally vulnerable and least capable of rational decision-making.

The Slippery Slope from 'Personalization' to 'Manipulation'

The EFF's report touches on a broader issue: the entire commercial model of online behavioral advertising is fundamentally flawed. When ad technology becomes powerful enough to predict and influence an individual's in-the-moment decisions, the line between "personalized service" and "psychological manipulation" becomes dangerously blurred.

For ordinary consumer goods, being served an ad for shoes you might like is largely harmless. But gambling is an inherently addictive domain capable of causing serious financial and psychological harm. In this context, AI precision targeting is no longer a convenience — it is a form of systematic exploitation of a known vulnerable population. Much of the Hacker News discussion centered on exactly this point: practitioners in the tech industry widely recognize that the same recommendation algorithms and user profiling techniques produce vastly different ethical outcomes depending on the industry they're applied to.

A pointed question emerges: do gambling platforms internally know which of their users are problem gamblers? If an AI system can detect addiction signals for marketing purposes, it is theoretically fully capable of detecting those same signals to intervene and protect users. Choosing the former over the latter reveals a value system in which commercial interests override user welfare.

Regulatory Lag and Technological Runaway

This case highlights how severely regulatory frameworks have fallen behind in the AI era. Existing gambling regulations largely target traditional advertising content and access controls — there is almost no effective oversight of the underlying data collection, algorithmic targeting, and predictive marketing that actually drive the harm.

The opacity of behavioral advertising makes regulation even harder. Users are typically unaware that they are being algorithmically "studied" and "locked in." The EFF has long advocated for stricter limits on behavioral targeting, and in some cases argues for an outright ban on surveillance-based advertising models. The DraftKings case provides an extreme but compelling piece of evidence for that position: when ad technology is used to amplify rather than mitigate social harm, industry self-regulation is clearly insufficient.

Potential remedies include: requiring gambling platforms to build addiction-detection and intervention mechanisms; restricting marketing directed at identified high-risk users; mandating transparency in algorithmic targeting logic; and limiting the collection and use of sensitive behavioral data at the source.

The Electronic Frontier Foundation (EFF), founded in 1990, is a US-based nonprofit digital rights advocacy organization with a long track record of speaking out on privacy, free speech, and technology regulation. Its core critique of behavioral advertising focuses on the "surveillance-based advertising" business model — the underlying logic by which platforms harvest user behavioral data at scale to build personal profiles, then sell access to those profiles as advertising inventory. EFF's preferred solution is not to patch the existing system but to cut it off at the source: ban cross-platform behavioral tracking and promote contextual advertising (ads served based on page content rather than user profiles) as an alternative. The DraftKings case is, for EFF, an extreme argument: it transforms the abstract harms of behavioral advertising into the concrete, targeted exploitation of an addicted population — and may help push legislators to bring algorithmic marketing under consumer protection frameworks.

A Mirror for AI Ethics

What makes DraftKings' practices so widely disturbing is not only that they involve gambling — a sensitive industry by any measure — but that they represent a paradigmatic example of the broader AI ethics dilemma. AI itself is a value-neutral technology, but when it is embedded in commercial systems with profit maximization as their sole objective, it can be directed to do what is most efficient — and most unethical.

This case is a reminder that when we discuss AI safety and ethics, we cannot focus exclusively on the distant risk of "superintelligence." We must also confront the concrete, present-day harms being driven by the technology that already exists. The ability to identify vulnerable people and exploit them has already been commercialized. The tech community, regulators, and the public all need to seriously ask: what, exactly, do we want AI to optimize for?

(This article is based on EFF's original reporting. The related discussion can be found in the Hacker News thread.)