How AI is rewiring modern sports betting

This week, the major league baseball betting scandal shook the sports world. Last month, all eyes were on the NBA. Why is sports betting in crisis? Does artificial intelligence worsen the odds for players or make sports betting safer? And what can gamblers do to protect themselves?

Sash Vaid is an associate professor of Marketing at the DeGroote School of Business. At the Vaid Lab for Dual-Use Technologies, Vaid and undergrad research assistant and Commerce student Audrey Zhu are investigating how marketing technology is rewiring modern gambling.

In a new research paper, Zhu, Vaid, and their co-authors Simona Liu (University of Toronto) and Daniel Gozman and Katina Michael (University of Sydney) discuss the current landscape of digital sports betting and explore socially responsible strategies for operators.

We asked Vaid and Zhu a few questions about their research and how gamblers can protect their best interests when the odds sometimes seem stacked against them.

“AI hasn’t made gambling fairer; it’s made it smarter, faster and far more asymmetrical in who truly holds control,” they told us.

How do the MLB and NBA betting scandals illustrate some of the dangers of new technologies in gambling?

The recent NBA and MLB betting scandals show how technology has made gambling both more sophisticated and more vulnerable to manipulation. In both cases, technology creates opportunities for insider information and micro-level rigging.

In the NBA case, players shared private details about injuries and game availability, which were then used to place highly specific “prop bets.” In the MLB case, pitchers were accused of altering the speed or outcome of individual pitches so bettors could profit from “pitch-level” wagers. These are markets that exist only because real-time tracking systems and algorithmic odds models make it possible to bet on single actions. When a single player can influence a one-off event, the integrity risk becomes enormous.

At the same time, both scandals revealed the use of advanced cheating technologies, from rigged card-shuffling machines and X-ray tables to specialized glasses that can read marked cards. These tools reflect a core idea in our paper: technologies originally developed for security, detection or optimization can be repurposed into instruments of manipulation. Their capabilities remain the same, but their function flips.

What’s alarming is that these systems make gambling look modern and controlled, when in fact they have made it harder to regulate. The scandals illustrate how optimization technologies can be weaponized when ethical boundaries are not enforced.

Thanks to AI, do gamblers face better or worse odds than they used to?

AI has changed what it means to gamble. It used to be a game of chance, spinning a wheel, rolling dice or betting on something unpredictable. Players use small algorithms or simple machine learning models to look for what are called “value bets”—moments when they believe the odds offered by the platform underestimate the real chance of winning. It gives them a sense of control and the belief that they can outsmart the odds through data and analysis.

However, the companies behind the platforms are running far more sophisticated systems that analyze millions of data points every second. These systems don’t just calculate odds or manage financial risk. They also track and analyze player behavior, everything from your betting frequency and preferred games to your emotional responses and withdrawal habits.

So, while individual players might use AI to make smarter bets, they’re also feeding the system more behavioral data that strengthens the operator’s predictive power. It’s a feedback loop: as you try to optimize your strategy, the platform’s AI optimizes its counter-strategy. In the short term, a few skilled players might benefit from their models. But in the long run, the house remains dominant with their algorithms, scale, computing power and behavioral insights. AI hasn’t made gambling fairer; it’s made it smarter, faster and far more asymmetrical in who truly holds control.

How is AI used in gambling to improve predictions about the outcomes of games?

AI now plays a major role in how modern gambling works, especially in sports betting. These systems pull in massive amounts of data, everything from team and player statistics to injury reports, weather conditions and even social media activity that reflects public sentiment. Machine learning models are then trained on all that data to predict how a game might play out: who’s likely to win, by how much and when momentum could shift. For example, some AI systems use patterns from thousands of past matches to estimate the probability of a certain team winning. They learn which variables matter most, how a player performs after travel or how a team reacts to pressure in overtime. The result is a prediction engine that can update odds in real time as new information comes in.

The gambling industry uses sophisticated tools and technologies. At what point do these techniques go beyond skilled marketing and enter an ethical grey area?

It crosses into an ethical grey area when personalization stops being about “helping users make decisions” and starts becoming a way to amplify emotional impulses, especially during high-arousal moments.

There are several red flags:

• When the platform knows your profile—your risk level, spending patterns, even when you’re most likely to keep playing—but you have no visibility or control over that data.
• When user experience experiments focus solely on increasing playtime or spending, with no attention to risk reduction or self-control.
• When systems identify and target high-risk or younger players because they respond more to rewards.
• When platforms use tiered memberships, variable rewards and instant feedback.

Once these features shift from behavioral marketing to behavioral manipulation, the issue is no longer marketing; it becomes a form of consumer control. Operators need an ethical bottom line, and profit can’t come at the cost of autonomy. Corporate social responsibility must be more than a checkbox—it must reflect a shared ethical commitment across stakeholders.

Dual-use technologies developed for defense have been repurposed in civilian markets, including online gambling platforms. Why is it important to understand the origins of dual-use technologies?

Understanding the origins of dual-use technologies—those designed for defense but repurposed for commercial use—is crucial because the intent behind their design shapes what they can do and the risks they carry. Systems designed for threat detection were originally developed for defense and intelligence purposes, to identify insider threats, suspicious activity or potential attacks. They now serve persuasive functions in consumer markets. These are classic examples of dual-use technologies, where defense-grade anomaly detection tools are repurposed for consumer behavior prediction.

When these tools migrate into the gambling world, their purpose shifts from protection to persuasion. They naturally evolved into player profile study, churn prediction and behavioral triggering.

Regulation needs to be risk-based, not label-based: it’s not about whether something is called “AI” but how it’s used and how it influences people. Recognizing that lineage also helps close the grey zones between innovation, surveillance and manipulation. This is especially true for dual-use AI systems, whose original defense applications may carry hidden risks when deployed in consumer-facing platforms.

What can gamblers do to protect themselves?

Self-protection starts with awareness. Most gambling platforms are built to be frictionless, to keep you engaged without ever feeling interrupted. To counter that, players can consciously add friction back into the system, by, for example, turning off notifications and delaying re-deposits.

Can regulations help reduce the misuse of AI technologies in gambling? In what ways could AI actually be used to improve guardrails in gambling?

We propose a dual-path approach: external regulation and internal guardrails, developed collaboratively across disciplines.

Externally, laws should treat AI systems according to their risk level. For example, the EU’s AI Act classifies systems by how much harm they can cause. Applying that logic to gambling means stricter oversight for technologies that personalize content, predict behavior, or influence decision-making. Operators should face independent audits that examine how algorithms are trained, what data they use, and whether players have been harmed as a result.

There should also be transparency requirements: users must know when personalization is happening, and they should have the right to turn it off. And safer defaults such as deposit limits, time caps and cooling-off periods should be standard, not hidden deep in settings menus.

Internally, AI can also be part of the solution. The same systems that track engagement can be reprogrammed to detect early signs of risky behavior: chasing losses, betting faster or playing longer than usual, and trigger intervention. Personalized “pause” or “withdrawal” messages can be just as data-driven as marketing prompts.

If you need help or support with problem gambling, you can call CAMH toll-free at 1-800-463-2338.