DraftKings Uses AI to Hunt Down and Target Losing Gamblers
DraftKings trained an AI model on its own user data to identify chronic losers and hit them with targeted ads to keep them betting.
Business · Source: Hacker News
What happened
DraftKings is using machine learning to identify customers who consistently lose money. According to a New York Times report, the sports betting company trains its AI on internal betting records to map out user behavior. Once the model flags a losing gambler, DraftKings hits them with targeted promotions. The explicit goal is to lure them back to the platform to place more bad bets. DraftKings knows that losing gamblers are the ones who actually drive their revenue.
The system relies entirely on first-party data. DraftKings does not need to buy external information from data brokers to make this work. They just feed their own users' behavioral history into the algorithm. The model learns exactly what kind of promotion will trigger a specific user to open their wallet again. Because AI operates as a black box, engineers are incentivized to hoard massive amounts of user data to keep refining the model's accuracy.
The Electronic Frontier Foundation is calling this out as a predatory practice. They point out that problem gamblers are the most vulnerable to this exact type of automated targeting. By re-engaging these individuals through targeted promotions, DraftKings capitalizes on their vulnerability for profit. The EFF argues this supercharges the harms of online behavioral advertising and proves that simply banning the sale of third-party data is not enough to protect consumers.
Key facts
- First-party data — DraftKings uses only data collected directly from users, not third-party brokers.
- The New York Times — The publication that originally reported on DraftKings' AI targeting practices.
Why it matters
This changes the baseline for how consumer apps use machine learning. We are moving past generic recommendation engines and into the era of automated financial extraction. Companies are now building highly specific models designed to exploit user vulnerabilities. If you have enough first-party data, you can predict exactly when a user is most likely to make a poor decision and serve them a customized trigger. This proves that you do not need third-party cookies or external data brokers to build a ruthless behavioral advertising engine. The data you collect inside your own app is more than enough to manipulate user behavior.
The second-order effect is a massive regulatory backlash against first-party data usage. Until now, privacy advocates and regulators focused heavily on stopping the sale of data to third parties. DraftKings proves that keeping data in-house does not prevent predatory behavior. Expect lawmakers to start looking closely at how internal AI models process user behavior. The EFF is already demanding a total ban on all behavioral advertising. Furthermore, government agencies like ICE are watching the ad tech space closely. They recently published a request for information to see how commercial big data can support their own investigations. The surveillance machine built for ads will inevitably be used by the state.
For builders
First-party data is your biggest asset
DraftKings built a highly effective targeting engine without buying outside data. Founders should focus on capturing deep behavioral events inside their own apps. The companies that own the proprietary data will own the most profitable models.
Expect new rules on internal AI models
Privacy laws currently focus on data sharing and third-party brokers. This will change soon. Builders should prepare for future regulations that dictate how you can use your own users' data to train internal machine learning models.
Black box models drive massive data hoarding
The EFF noted that engineers cannot always predict which data points the AI finds useful. This creates an incentive to collect everything. Founders must balance the need for model accuracy against the growing liability of storing massive amounts of user data.
My take
DraftKings is doing exactly what they are incentivized to do. If you build a business model on users losing money, your AI will optimize for finding losers. We can complain about ethics all day, but this is just raw capitalism executing code. If you want to stop it, you have to change the rules of the game. You cannot ask an algorithm to be nice when the financial incentives demand ruthless efficiency.
Original reporting: Hacker News. This is my rewrite and opinion.