INTERVIEW JUNE 2026
Fragmented by Design: What iGaming Operators Get Wrong About AI, Fraud, and the Player They Share
SEON's CEO built a fraud prevention company after being defrauded himself. He still thinks the industry is solving the wrong problem.
Interview
Our editor had a conversation with Tamás Kádár, CEO and Co-founder of SEON, about why near-universal AI adoption has not solved the fraud problem — and why the real bottleneck has never been the technology. Ninety-eight per cent of iGaming operators now use AI in their fraud and AML workflows. Ninety-five per cent say it works. Headcount is still rising. Budgets are still rising. If that reads like a contradiction, Tamás Kádár says it is — and it is the most important thing the industry is not talking about..
Tamás Kádár, CEO and Co-founder, SEON. Image courtesy SEON
Box-out: Tamás Kádár, CEO and Co-founder, SEON
Current Title: Senior Director of Communications
Company: SEON
Background: Tamás Kádár co-founded SEON in 2017 with Bence Jendruszák. Both were students at Corvinus University in Budapest, where Kádár studied Deep Information Communications, when they began building a crypto exchange for the Central and Eastern European region. The platform was targeted by fraudsters within its first week of accepting card payments, and no adequate tool existed to protect against it. They built one themselves. SEON is headquartered in Austin, Texas, with 100 employees in the United States, approximately 80 of them in Austin. The company also has 200 employees in Budapest and 30 in London (and more than 350 worldwide). Clients include Flutter Group, Entain, Evoke, and Bally's. Kádár has been recognised in Deloitte's Fast 50 and writes for the Forbes Technology Council and HackerNoon.
Kádár founded SEON after being defrauded himself, and the company has spent nearly a decade helping operators detect financial crime, verify identities, and stay compliant in real time. SEON works with Flutter Group, Entain, Evoke, and Bally's, among others. Its edge, as Kádár describes it, is not the capability of any individual tool. It is the ability to connect signals across the entire customer journey — from onboarding through login, deposit, withdrawal, and everything in between. The gap the industry has, he argues, is not a shortage of AI. It is a shortage of connected intelligence.
That distinction — between having AI and having AI that talks to itself — sits at the centre of every significant question now facing operators trying to manage fraud, AML, and responsible gambling simultaneously.
Industry perspective
"The bottleneck is no longer whether AI works. It's everything around it: disconnected data, siloed teams, slow implementations."
— Tamás Kádár, CEO & Co-founder, SEON
The Paradox Nobody Is Talking About
Kádár's framing of the AI paradox is precise: AI is not replacing analysts. It is making them more effective at dealing with problems that are themselves growing more complex. Fraud networks have become more sophisticated. Cross-channel activity has increased. Financial crime is not standing still. And as operators improve on the metrics they are measured against — detection rates, alert resolution times, case closure — they keep uncovering more complexity underneath.
The assumption that AI would reduce the need for people, he says, has simply not held. What it has done is change what those people spend their time on. The strongest teams are using AI to remove the repetitive work — pulling data from systems, copy-pasting between tools, assembling a complete picture of a customer from multiple sources that do not communicate with each other. That frees analysts to focus on the cases that require genuine human judgment: the edge cases, the cross-channel fraud rings, the situations where the pattern does not fit any established algorithm.
"What we are seeing is more sophisticated fraud networks, more cross-channel activity, and more cases that require human judgment at the same time," Kádár says. "AI is not going to replace the analyst. What it is going to help with — what it is already helping with — is turning the time for analysis into a smaller piece, so they can be significantly more effective working on one case."
The structural problem underneath the paradox is fragmentation. An analyst in the fraud department may be using an entirely different system from a colleague in compliance. Someone in the responsible gambling team may be using a CRM or a third tool that has no connection to either. There is no shortage of signals. There is a shortage of those signals reaching the right person, in the right context, at the right time. The overhead of bridging that fragmentation — the communication, the manual data transfers, the duplicate case management — is where the time goes.
"Operators don't have a shortage of signals," Kádár says. "What they have a shortage of is connected intelligence." SEON is building toward what he calls an AI command centre: a single layer that collects risk-related signals across the entire customer journey and makes them available to every function that needs them. The direction the industry is heading, he argues, is consolidation — fewer stack components, fewer vendors, fewer gaps between them.
SEON AML multi-signal risk view. Image courtesy of SEON
The Data Converges. The Organisation Does Not.
The question of whether fraud prevention and responsible gambling are converging at the data layer is one Kádár answers carefully. The data, he says, is converging faster than the organisations are.
The same behavioural patterns — sudden changes in activity, unusual funding patterns, escalating spend, behaviour that simply does not look right — appear across multiple risk domains simultaneously. They do not sit neatly in one department's remit. A customer flagged for unusual deposit behaviour may be a money laundering risk, a problem gambler, or both. The signal is the same. The team that sees it, and what they do with it, depends on how the operator has structured its internal functions and which system that analyst happens to be using at that moment.
"Historically, fraud, AML, and responsible gambling evolved as separate functions because they had different regulatory drivers and different internal ownership," Kádár says. "But the customer doesn't care about our structures or our departments." The operators getting ahead, in his view, are those building a more holistic view of customer risk across the entire journey — asking not which department owns a particular signal, but what a complete picture of that customer's behaviour actually shows.
SEON is expanding its product offering in direct response to this convergence. A session monitoring product currently in beta is designed to track customer behaviour across the platform in granular detail: time spent on each game, buttons clicked, how long it takes to fill each form, what is entered in each field. The product is intended to help identify account takeover attempts, which have become, according to SEON's clients, one of the primary concerns of the past twelve months. Illicit activity, operators are telling SEON, has shifted from the onboarding stage — where KYC and identity verification have become more robust — to the post-onboarding stage, where social engineering, credential theft, and account access are the real vectors.
SEON is also developing a responsible gambling layer that would extend this behavioural monitoring to surface signals relevant to player harm — though that product has not yet been released. What launched in June 2026 is different in focus but related in architecture: an MCP server that allows analysts to connect their preferred AI tool — Claude, ChatGPT, Gemini, Microsoft Copilot, or any custom agent — directly to SEON's real-time risk signals covering identity, device, behaviour, AML, and IP data. The logic is straightforward: the data pipeline, not the AI capability, has been the bottleneck. The new server removes the need for analysts to work inside SEON's own dashboard to access its data. "The software world is moving toward a headless model, where teams don't need to live inside a vendor's dashboard to get full control over data and functionality," Kádár said at launch. The analyst, in his framing, becomes the designer — setting the parameters, choosing the tools, and supervising what the system does with them.
Compliance as Strategy, Not Administration
On the question of whether tightening regulation in the UK, the Netherlands, and Sweden is producing genuinely better detection, or primarily better documentation, Kádár is direct: regulation raises the floor. It does not automatically create excellence.
The difference, he argues, comes down to mindset. Operators who treat regulatory change as a compliance exercise — who ask what they need to do to pass an audit and then move on — tend to end up with better reports and similar risk exposure. Operators who use regulatory change as an opportunity to improve how they actually manage risk tend to end up as stronger businesses. The question the best organisations ask is not what the regulator requires. It is what they need to know in order to make better decisions and protect the business.
"Compliance becomes strategic rather than administrative when an organisation stops asking what it needs to do to satisfy an audit and starts asking what it needs to know to make better decisions," Kádár says. "The operators that see player protection and compliance as part of the same strategy are the ones that will be stronger over the long term."
SEON does help with the documentation side. The platform includes a component that uses AI to automatically populate suspicious activity reports — drawing on data already held in the system and applying standard FinCEN templates where required. The system drafts the report; no analyst needs to write it up manually. Kádár is clear that this is an after-step, a reporting obligation that follows a decision already taken. The harder work is upstream: the real-time algorithm that runs on every deposit, withdrawal, and login, and the operator's willingness to act on what it surfaces.
SEON's unified AI data layer — a network of risk signals spanning device intelligence, IP data, AML flags, and behavioural indicators, feeding directly into any connected AI tool via open infrastructure. Image courtesy of SEON
Building for the Next Stage
Kádár has described the same vision for SEON since the company started nine and a half years ago: to become the category leader in fraud and compliance infrastructure — the go-to solution for any online business accepting deposits and managing risk. That ambition has not changed. What has changed is the scale from which he is pursuing it.
SEON now employs over 350 people worldwide across Austin, Budapest,London, and Singapore, and has grown from a two-person founding team into a company whose clients include some of the largest operators in the world. For Kádár, the question driving the next phase is not scale for its own sake. It is product prioritisation: with more ideas than any team can build, choosing the right two from any given ten.
The appointment of Sagnik Nandy — CTO and EVP of Engineering at DocuSign, who previously held senior roles at Google and Okta — to SEON's board in April 2026 was made with exactly that problem in mind. Kádár describes Nandy as a sounding board for navigating product decisions at a stage of growth where the range of possibilities is wider than the resources available to pursue them. "We see great productivity gains over the last six to twelve months by adopting AI tools," Kádár says. "What's more challenging is finding out which two of ten ideas are the best ones — and then making sure they are well scoped with the help of our clients."
The aspiration is to be present in every Request for Proposal (RFP) and Request for Information (RFI) from iGaming and fintech operators within the next two to three years — not just as a fraud tool, but as the primary compliance infrastructure layer. That requires improvements in brand visibility, product breadth, and the quality of service delivered to existing clients. The category leader ambition, in other words, is not a marketing position. It is an operational target.
Kádár began building what would become SEON while still a student at Corvinus University, having lost a few thousand euros to fraud in the first week of running a crypto exchange. The company became a full-time pursuit after he left. He describes the experience of building from that point — with no corporate experience, no startup experience, and no prior knowledge of the fraud prevention space — as a process of becoming subject matter experts by necessity. "I was a fresh grad when we started SEON," he says. "No corporate experience, no startup experience. The entire journey has been the learning curve. I can't point to one lesson because I didn't know anything when we began. What's made the difference is surrounding ourselves with people who are smarter than we are. That's been key from day one."
The problem that started SEON — inadequate tooling for operators trying to protect against fraud in real time — is, in his view, still far from solved. The company has addressed parts of it. The fragmentation that keeps operators from connecting the tools they already have remains. That, more than any product launch or board appointment, is what still drives the business.
Written by Elisabet Johansson.
Further Reading & Key Sources
SEON, AI Reality Check: 2026 Fraud & AML Leaders Report, February 2026
SEON: Global iGaming Risk & Fraud Report, March 2026
SEON: MCP server and AI capabilities launch, 2 June 2026
© 2026 iGaming Review | Provider: iGaming Review International AB | Registered in Sweden | Legal Notice
