How DraftKings Uses AI to Target Likely Losing Bettors

Introduction

Artificial intelligence is changing how sports-betting companies identify, rank, and communicate with customers. A New York Times report, as described in several social-media posts, alleges that DraftKings uses AI and data science to identify bettors considered most likely to lose and target them with betting incentives. The reporting also reportedly raised questions about why similar technology was not used as aggressively to identify customers at risk of gambling harm. Source 1 Source 3

The allegations matter because betting platforms collect extensive behavioral information. They may track how often customers bet, which sports they select, how much they deposit, how they respond to promotions, and whether they continue betting after losses. Predictive systems can turn those observations into estimates of what a customer may do next.

This technology can support legitimate personalization. It can also create a serious consumer-protection problem if the people most likely to generate revenue are also those showing signs of financial vulnerability or compulsive betting.

The central question is not whether AI can predict betting behavior. It is how companies use those predictions. A system that identifies a customer as likely to continue betting and lose money could restrict promotions, offer support, or encourage a cooling-off period. It could also send more attractive incentives.

What the Report Allegedly Revealed

The New York Times report described in the cited posts allegedly says that DraftKings uses AI and data science to identify customers most likely to lose and target them with betting incentives. Source 3 Other posts similarly describe an investigation into DraftKings’ use of AI to target bettors likely to lose. Source 7 Source 9

“Likely to lose” does not necessarily mean that a customer has a gambling disorder or has already experienced serious harm. The phrase could describe customers who lose occasionally, lose consistently, bet frequently without withdrawing funds, increase wagers after losses, or display behavior associated with financial or psychological risk.

These categories overlap but are not identical. A recreational bettor can lose money without displaying harmful behavior. Conversely, a customer may show risky patterns before suffering visible financial damage.

The available source summaries do not establish the precise model design, data inputs, or customer classifications used by DraftKings. Those details require confirmation from the original reporting, company disclosures, regulators, or DraftKings itself.

How AI Can Target Sports Bettors

A betting platform may analyze:

• Bets placed and markets selected.
• Betting frequency and wager size.
• Deposit and withdrawal patterns.
• Session timing and duration.
• Responses to previous promotions.
• Changes in betting amounts.
• Betting after wins or losses.
• Devices, locations, and access channels.

A simplified targeting process may involve four steps:

1. The platform gathers historical behavioral data.
2. A model identifies patterns associated with future betting, retention, or losses.
3. Customers are assigned to predictive segments.
4. Marketing systems use those segments to select offers or messages.

AI does not need to understand why a customer is betting. It only needs to estimate what that customer is likely to do next.

Three concepts should be distinguished:

• Prediction: Estimating future behavior.
• Personalization: Tailoring an offer or message.
• Intervention: Attempting to reduce harmful behavior.

The same signal may support all three functions. Repeated deposits could predict continued betting, trigger a personalized promotion, or prompt a responsible-gambling review. The ethical outcome depends on which action the company chooses.

Why Likely Losing Bettors May Be Valuable Targets

Sportsbooks generate revenue through the margin built into odds, customer losses, fees, and related products, depending on the market and outcome. An individual customer loss does not automatically equal direct company profit on every bet.

Over time, however, customer activity and expected losses can influence customer-lifetime-value calculations. A bettor who remains active, places frequent wagers, and deposits repeatedly may be considered valuable even when individual bets produce different outcomes.

Promotions can also change how customers perceive risk. A bonus may make another bet feel cheaper, a limited-time offer may create urgency, and a loss-recovery promotion may encourage chasing. A personalized message can make continued betting feel rewarding.

The concern is strongest when incentives are directed at people predicted to continue betting and lose. Personalization may then increase exposure to the behavior the system has identified as commercially valuable.

A responsible system should distinguish ordinary commercial activity from potential harm before converting behavioral indicators into marketing decisions. Possible safeguards include affordability checks, spending limits, cooling-off periods, transparent promotion rules, and restrictions on incentives after loss-chasing behavior.

The Responsible-Gambling Contradiction

Signals associated with gambling risk may include rapid increases in betting, repeated deposits, escalating wager sizes, betting soon after losses, unusually long sessions, attempts to recover money quickly, and repeated use of promotional funds.

None of these signals proves addiction. A customer may place a large wager because of a one-time event or bet more frequently during a major tournament. Risk detection requires context, multiple indicators, customer communication, and human review.

Even so, the overlap between commercial targeting and harm detection creates an obvious policy question: if a system can identify customers likely to continue betting and lose, why should it not also identify customers who may need protection?

The reported criticism focuses on an alleged imbalance. Predictive technology may be used to find customers likely to remain active and lose, while comparable technology is reportedly not applied as forcefully to restrict promotions or interrupt harmful behavior. Source 3

Commercial incentives can favor continued betting. Responsible-gambling systems can produce the opposite result by reducing deposits, limiting promotional eligibility, or encouraging customers to stop. Effective intervention may reduce short-term revenue, but consumer protection requires accepting that possibility.

AI also has limitations. False positives can incorrectly flag customers, false negatives can miss people at risk, historical data can reproduce bias, and automated decisions may be difficult for customers to understand. AI should support—not replace—responsible-gambling teams, trained support staff, and independent oversight.

Risks for Young and Vulnerable Customers

The cited public reaction raised concerns about teenagers. Source 5 The available summaries do not establish that DraftKings specifically targeted minors, and that claim should not be made without reliable evidence.

Young people may nevertheless face greater exposure to sports-betting advertising through broadcasts, mobile applications, social media, influencers, and peer networks. They may have less experience evaluating probability, rewards, and long-term financial consequences.

Customers experiencing debt, unstable income, gambling-related stress, previous losses, or compulsive betting patterns may be particularly vulnerable to incentives. A promotion can be harmful when it reaches someone already trying to recover losses by encouraging another deposit or extending a betting session.

Customers showing vulnerability require stronger protections, not more aggressive incentives. Any system that identifies risk should consider promotion restrictions, support messages, spending controls, and human review.

Data privacy adds another concern. Customers may reasonably ask what data the platform collects, how it infers the likelihood of future losses, whether profiling is clearly disclosed, whether personalized promotions can be refused, how long behavioral data is retained, and who can access the predictions.

Regulatory and Policy Questions

Customers may deserve clear notice that AI-based profiling is used to select promotions. Meaningful disclosure could explain which data categories influence marketing, whether loss patterns affect promotional eligibility, whether high-risk customers receive different offers, how to opt out, and how to challenge an automated decision.

Regulators could consider prohibiting promotions after signs of harmful betting, requiring affordability assessments, banning loss-chasing incentives, limiting personalized promotional messages, requiring cooling-off periods, auditing models for consumer harm, and requiring records of promotion-related complaints and losses.

They could also require early-warning systems, human review of high-risk accounts, automatic promotion restrictions, customer spending and time limits, independent audits, and public reporting on intervention outcomes.

The question of whether Ontario Premier Doug Ford would respond illustrates the broader policy debate, not an established government action. Source 5 Any assessment of Ontario’s response should verify statements from provincial officials, regulatory actions, advertising rules, responsible-gambling requirements, investigations, and legislative proposals.

What Betting Platforms Should Explain

Companies should provide plain-language information about their models’ objectives, the data categories used, customer segmentation, promotion selection, human oversight, and limits on automated decisions. Proprietary technology does not eliminate the need for accountability.

Companies should also disclose whether customers showing loss-chasing or escalating behavior receive fewer promotions, responsible-gambling messages, contact from trained staff, automatic restrictions, or referrals to limits, cooling-off periods, and self-exclusion.

Independent audits should evaluate false positives, false negatives, disparate effects, customer complaints, intervention outcomes, and promotion-related losses. Model performance should be measured by reduced harm and meaningful customer outcomes, not merely by the number of alerts generated.

How Customers Can Protect Themselves

Customers can disable marketing notifications where possible, unsubscribe from promotional email and SMS messages, avoid acting on time-limited offers, and read promotion terms before accepting an incentive.

Depending on local availability, platform controls may include deposit limits, betting limits, time limits, session reminders, cooling-off periods, and self-exclusion. Limits work best when set before betting becomes difficult to control. Customers should avoid increasing limits during stress or after a loss.

Support may be appropriate when betting causes debt, relationship conflict, anxiety, concealment, missed work or school, or repeated attempts to recover losses. Gambling-support services, financial counselors, and mental-health professionals can provide assistance.

Conclusion

AI can help betting platforms predict who will keep gambling and who may lose. The ethical issue is how companies use those predictions.

Targeting likely losing bettors with incentives creates a conflict between revenue optimization and consumer protection, especially when the same behavioral signals may indicate gambling-related harm. The allegations summarized in the cited New York Times coverage deserve careful verification, transparent responses, and regulatory scrutiny. Source 1 Source 9

Responsible AI in sports betting requires transparency, limits on personalized promotions, early intervention, independent audits, and stronger protections for young and vulnerable customers.

Predictive technology should not identify vulnerability only when vulnerability creates a commercial opportunity.

FAQ

Does DraftKings use AI to target bettors likely to lose?

The New York Times report summarized by the cited social-media posts alleges that DraftKings uses AI and data science to identify bettors considered most likely to lose and target them with betting incentives. The precise model design and targeting criteria require confirmation from DraftKings or the original report.

How can AI predict which bettors are likely to lose?

AI systems can analyze betting frequency, wager size, deposits, withdrawals, selected sports and markets, session behavior, and responses to promotions. These patterns may help predict future activity or losses, but they do not prove that a customer has a gambling disorder.

Why are incentives for likely losing bettors controversial?

Incentives may encourage additional deposits, more frequent betting, longer sessions, or attempts to recover previous losses. The concern is stronger when the same behavior used for targeting could indicate financial vulnerability or gambling-related harm.

Could DraftKings use AI to identify gambling-risk behavior?

Similar behavioral signals could potentially help identify customers who need support. AI cannot diagnose addiction by itself. Effective protection requires context, human review, transparent safeguards, and access to limits or self-exclusion.

What can customers do to reduce targeted betting promotions?

Customers can disable marketing notifications, unsubscribe from promotional communications, set deposit or betting limits, use cooling-off tools, and request self-exclusion where available. They should avoid chasing losses and seek independent support when betting causes financial or personal harm.

What should regulators investigate?

Regulators could examine how platforms profile customers, whether high-risk bettors receive targeted incentives, how models are audited, what disclosures customers receive, and whether responsible-gambling systems restrict promotions when harmful behavior appears.