Kalshi partners with smaller rival Novig to ask 35 states for access to "self-exclusion lists", which identify gambling addicts, to restrict problem gamblers (Bobby Allyn/Bloomberg)

Kalshi and Novig Want Self-Exclusion Lists in 35 States: What That Means for AI, Privacy, and Responsible Gambling

By · Published November 19, 2026 · Updated November 19, 2026

Kalshi and its smaller rival Novig are asking 35 states for access to self-exclusion lists. Those lists identify gambling addicts. The goal is to restrict problem gamblers from using the companies’ services. On the surface, it is a responsible-gambling request. Below the surface, it is a major test case for how AI will be used to protect people, verify identities, and handle sensitive data in regulated markets.

The request matters because self-exclusion lists are one of the few tools that turn a person’s own decision to stop gambling into an enforceable block. If Kalshi and Novig get access, they will need a way to check those lists at sign-up, at login, and as accounts change. That is exactly the kind of high-volume, high-stakes matching problem that AI systems are being built to solve. It is also the kind of problem where a mistake can harm a vulnerable person or wrongly block someone else.

The 35-State Request, Explained

Kalshi and Novig are competitors. Kalshi is the larger player, and Novig is the smaller rival. They are working together to ask 35 states for access to self-exclusion lists. The lists are maintained by states and are meant to identify gambling addicts. When a person joins a self-exclusion list, they are saying they want to be blocked from gambling. The list helps operators know who should not be allowed in.

The request is simple to describe. The companies want the data so they can restrict problem gamblers. The hard part is everything that comes after access is granted: how the data is stored, how identities are matched, how false positives are handled, and how the system proves it is not being misused.

This is not just a policy story. It is a preview of how AI will sit between public health rules and private companies. The companies that get this right will build trust. The companies that get it wrong will face lawsuits, fines, and reputational damage.

Why Self-Exclusion Lists Matter

A self-exclusion list is a registry. A person asks to be added, and their name is shared with gambling operators so they can be blocked. For a person with a gambling problem, that block can be a lifeline. It removes the ability to act on an impulse. It creates friction at the exact moment when friction is needed.

But a list is only useful if operators actually check it. Manual checks do not scale. A person can create a new account with a different email address, a different phone number, or a slightly different name. A list check that only happens once at sign-up can be bypassed. That is where AI comes in.

AI systems can match identities across records. They can compare names, dates of birth, addresses, device fingerprints, and other signals. They can flag a new account that looks like a blocked person. They can also help operators avoid blocking the wrong person, which is a real risk when names are common or when data is incomplete.

The AI Layer: Matching, Monitoring, and Privacy

The first AI use case is identity matching. Kalshi and Novig would need to compare the people who use their services against the state lists. That sounds like a simple database lookup, but it is not. State lists may have different formats. Names may be spelled differently. People may move. A person may use a nickname. AI can help normalize the data and find likely matches.

The second use case is continuous monitoring. A person who is blocked at sign-up may try again later. AI can watch for new accounts that share signals with a blocked identity. It can also watch for changes to an existing account, such as a new payment method or a new device, that suggest a blocked person is trying to get back in.

The third use case is privacy-preserving matching. Companies do not need to see the full list to check it. AI and cryptography can allow a company to ask a state system a yes-or-no question without exposing the whole list. Techniques like hashing, tokenization, and secure matching can reduce the amount of sensitive data that moves between systems. That matters because self-exclusion lists contain health-related information about gambling addiction.

The fourth use case is compliance automation. States will want proof that companies are checking the lists. AI can generate audit logs, flag missed checks, and produce reports. It can also help companies respond when a state changes its rules. That is a back-office use case, but it is just as important as the customer-facing one.

Responsible Gambling Is Becoming an AI Product Category

For years, responsible gambling was mostly a policy page and a hotline number. That is changing. Regulators want proof that operators are actively reducing harm. Self-exclusion lists are one part of that. AI-driven risk detection is another.

AI can look for patterns that suggest a person is losing control. It can flag rapid deposits, long sessions, chasing losses, or a sudden change in behavior. It can trigger a cool-off period, a deposit limit, or a human check-in. When a person is on a self-exclusion list, the AI should not need to guess. The list is an explicit signal. The system should block the person and log the event.

The Kalshi and Novig request shows that competitors can cooperate on safety. That is a big shift. Safety is often treated as a cost center. But in regulated markets, safety can become a shared utility. If both companies use the same list-checking standard, the whole market becomes harder to exploit. AI makes that shared utility practical because it can handle the volume and the complexity.

The Privacy Trap: Why Access Is Not the Same as Permission

Access to a self-exclusion list is not the same as permission to use it however you want. The data is sensitive. It identifies people with a gambling addiction. If it leaks, people could face shame, discrimination, or worse. If it is used for marketing, it becomes a targeting tool. If it is shared too widely, it becomes a surveillance database.

AI raises the stakes. A model trained on self-exclusion data could learn patterns that are not about gambling at all. It could pick up on income, location, or other traits. That is why governance matters. Companies need purpose limitation: the data is used only to block problem gamblers, not to score customers or build profiles. They need data minimization: collect only what is needed for the match. They need audit trails: every check should be logged and reviewable. They need appeal processes: a person who is wrongly blocked should be able to fix it.

There is also a fairness problem. AI matching can produce false positives. A person with a common name might be blocked because they share a name with someone on the list. A person who moved might be flagged because an old address matches. These errors are not just inconvenient. They can feel like an accusation. Companies need human review for close calls and a clear way for people to challenge a block.

What This Means for Businesses and Regulators

For businesses in regulated markets, the message is clear: plan for an AI compliance layer. If you operate in a state that offers self-exclusion, you will likely need to check it. If you operate in 35 states, you will need a system that can handle different formats and different rules. Doing that manually is not realistic. AI can help, but only if it is built with privacy and fairness in mind.

For regulators, the message is also clear: access is not enough. States should set standards for how lists are shared, how matches are confirmed, and how errors are corrected. They should require auditability. They should fund privacy-preserving technology. They should also watch for mission creep. A system built to block problem gamblers should not become a general-purpose identity surveillance tool.

For AI builders, this is a product opportunity. Responsible gambling is a growing category. The winners will be the companies that can prove their systems are accurate, private, and fair. That means building explainable matching, bias testing, consent management, and secure data sharing from day one.

The Road Ahead: Shared Safety Utilities

The Kalshi and Novig request points to a future where safety data is shared through standard APIs. A company would send a match request, and a state system would return a yes or no. The company would not need to store the full list. The state would not need to trust every company with a copy. AI would handle the matching in the middle.

That future is not guaranteed. It depends on trust. If companies misuse the data, states will pull back. If AI produces too many errors, the public will lose confidence. If privacy is an afterthought, the system will face legal challenges. The companies that treat this as a public health project, not a data grab, will be the ones that succeed.

The larger lesson is about AI’s role in regulated markets. AI is often framed as a way to grow revenue. Here it is a way to enforce a safety rule. That is a different kind of use case. It requires different metrics: not clicks or conversions, but blocked accounts, reduced harm, and fair outcomes. It also requires different governance: not just model accuracy, but privacy, auditability, and appeal rights.

Actionable Insights

Conclusion

Kalshi and Novig’s request to 35 states is a small headline with a big meaning. It shows that self-exclusion lists are becoming part of the operating stack for gambling-adjacent companies. It also shows that AI will be the tool that makes list checks practical at scale. The companies that get the balance right, accurate matching, strong privacy, fair appeals, will build trust with regulators and users. The companies that rush will create harm and invite backlash.

For the rest of us, this is a preview. The same pattern will show up in other sensitive areas: age verification, fraud checks, sanctions screening, and health data. AI will be asked to enforce rules that protect people. The question is not whether AI can do it. The question is whether we can build it with the guardrails that sensitive data demands. Kalshi and Novig’s 35-state request is the first test. It will not be the last.

TLDR: Kalshi and smaller rival Novig are asking 35 states for access to self-exclusion lists so they can block problem gamblers. The request is a responsible-gambling move, but it is also a major AI test case. AI will be used to match identities, monitor for new accounts, automate compliance, and preserve privacy. The winners will be companies that build accurate, private, and fair systems with audit trails and appeal processes. The losers will be companies that treat sensitive addiction data as just another dataset. This 35-state push shows how AI will be used to enforce safety rules in regulated markets, and why governance matters as much as accuracy.