The gambling hub DraftKings could be using the mountains of user data it collects, and the analyst expertise it boasts, to stop problem gamblers from ruining their lives.
Instead, it’s using this power to do the opposite, according to the New York Times: scout for potential addicts and keep them hooked.
To do that, DraftKings, which is best known for its sports betting app, built an AI model to determine which customers were likely to gamble and lose the most money in response to its enticing promotions, according to numerous former employees, some of whom helped build the machining learning algorithm. It predicts which gamblers could lose the most, scoring them based on a trove of their personal habits. A high score meant that a gambler was more likely to lose more money on each promotion DraftKings offered to keep them gambling.
Jayden Butts, a former data analyst at the company who tested the AI model, described the initiative to the NYT.
“We are looking for traits and features that we can target that indicate a good investment,” he told the newspaper. By strict financial logic, the “best investment would be a problem gambler.”
When you’re a chronic gambler, you only have so much money to lose. DraftKings, though, tried to bring users back into the fold by bombarding them with promotions that promised “free” bets, a deposit bonus, or a “profit boost.”
Blasting these promotions out randomly isn’t a smart strategy. That’s where the AI model comes in: to find the biggest gambling addicts who’ll fall for these offers hook, line, and sinker.
“It is as predatory as it sounds,” a different former data analyst who quit DraftKings in 2024 told the NYT. “If you lose more, we give you more, so you keep playing more.”
DraftKings began its efforts to build a model predicting how promotions would translate into a customer’s wins or losses in mid-2023, per the reporting. The driving question behind it, according to Butts, was whether “this person going to give us more than we’re giving them?”
“And if the answer is yes,” he added, “open the floodgates.”
The level of data surveillance, ethics aside, is impressive. Documents showed that DraftKings analyzed how frequently gamblers played, their daily account balances, and compared how much they lost to how much they bet. It also used another model that predicted how likely a customer was to stop gambling. One employee said he worked on a system that identified customers who were at risk of quitting the platform and tried bring them back.
That trove of data could also be put to more ethical means: finding problem gamblers and intervening.
But employees who tried to develop a model to help addicts say their efforts were quashed. One of them, Nestor Hernandez, started building a system to find vulnerable gamblers and assigning them a risk score, which the company could use to warn or even ban the customer before their losses spiraled. But the project was canned in 2025 just months after he left the company. Jake Shannin, who stayed working on the project until it was shut down, said the team was scheduled to present their model to DraftKing’s chief responsible gaming officer Lori Kalani, but their presentation was suddenly cancelled on the day it was supposed to take place.
DraftKings, in a statement to the NYT, rejected “any implication that its marketing practices are unfair or improperly targets customers.” Kalani, meanwhile, insisted to the paper that the company leadership made a “collective decision” not to use predictive models for identifying problem gambling because the technology “wasn’t evidence-based.”
More on gambling: Sports Betting Site Boasts That It’s Letting Users Bet on Little League Players