The payments, fintech and financial services leaders who filled the Venetian for this year’s Money20/20 USA arrived with various product offerings, different regulators and unique fraud problems. By the end of the week, their language coalesced around one pressing issue: the fraud controls many of the institutions rely on were built for a business they’ve already outgrown.
That gap surfaced across stablecoins and agent-driven payments that institutions are trying to navigate securely, in product roadmaps outrunning current fraud programs, in identity checks faltering with genAI-produced synthetic identities at scale, in thin-file applicants treated as a risk instead of a revenue opportunity and in AI budgets pointed at a sliver of the customer. Here’s what each one of these gaps means for the teams in the room.
The Product Roadmaps Unsupported by Fraud Programs
Companies at Money20/20 reported that they are outgrowing their fraud stacks built for initial offerings. A digital wallet that added lending, or a BNPL provider that started issuing cards, is running new products on controls designed for the old ones, and every addition widens the gap between what the business does and what its fraud stack can handle.
Nowhere was this clearer than in the week’s recurring debate over whether fintechs should become banks. On the “To Buy/Not Buy a Bank” stage, operators walked through what achieving chartered status actually costs: one fintech leader described a charter pursuit running two to two and a half years, restructured equity and voting rights, four bankers appointed to the board and an entire banking risk organization built from scratch — and the charter still wasn’t approved.
The consistent message was that while getting a license is attainable, operating under it is where the toughest demands begin. Panelists kept returning to the point that the compliance and fraud infrastructure a company builds for its first product is never the infrastructure its later products need, and that the gap becomes visible once a regulator or a loss exposes it.
That kind of tension was present throughout the event, surfacing in banking-sponsored sessions, where the argument centered on how governance and oversight must be thoughtfully considered rather than bolted on after the exam, and the topic was implicit in every conversation about launching into a new market or product line. The question the floor kept circling was how to get fraud teams into the room while a product is being planned, rather than after the first loss lands.
The Thin-File Gap Shapes Fraud Approval Decisions
Tradeshow conversations on the lending track moved past the familiar call for more data to a harder question: Which data actually separates a real first-time borrower from a fabricated one? Risk teams were clear about what they wanted — more insight into each applicant, without giving up control of the decision. The signals that answer that question rarely appear in a traditional data: the age of an email address, the history behind a phone number, whether a real adult’s life shows up online at all.
Whether or not attendees used the term, they were describing digital footprint analysis: reading the trail a person leaves across email, phone and the open web to tell an established identity from a manufactured one. Those signals already exist. The lenders closing the thin-file gap are the ones reading them.
When a Non-Human Actor Moves Money, Who Carries the Risk?
Stablecoins and agentic commerce dominated the agenda by sheer volume of sessions, with stablecoin infrastructure filling every day’s schedule and well over a hundred of the event’s sessions touching agentic AI. For fraud and risk teams working with stablecoins, two themes kept coming up: when a non-human actor moves money across a blockchain rail, who verifies the transaction, who carries the liability and where is the audit trail?
It also remains unclear who will capture the revenue once money routinely moves on-chain. While infrastructure is being built in the open, companies will continue tripping on conversions between stablecoins and fiat, with several speakers expecting that friction to fade as adoption deepens.
Agentic commerce carried the same unresolved questions with know your agent (KYA) as the dominant theme, even though nobody could point to a framework that enforces it yet. Mastercard, Stripe, J.P. Morgan and Google all took the stage on what agent-initiated payments will demand, and the gap underneath every session was the one running through this whole recap: the capability is shipping faster than the means to secure it.
Fraud Scams Are Still a Massive Issue
Several risk leaders described a specific use case more frequently cropping up, in which a genuine applicant, using their own documents on their own device, completes an application while someone on the phone dictates every answer. We call it coached fraud, and it frequently defeats identity verification for the simple reason that the identity is real.
The only trace it leaves is in how the application is filled out — long pauses while instructions are relayed, answers pasted rather than typed, a screen-sharing tool quietly running in the background. Session monitoring reads those behaviors as they happen, which is why it drew more questions than anything else at SEON’s booth.
This type of scam illustrates one example of a broader move toward attacks that can pass a point-in-time check. Synthetic identities assembled by AI, income documents generated on demand and deepfakes that defeat selfie checks are the fraud trends defining 2026, and each one beats a single moment of verification.
Where This Leaves the Industry
The conversations and agenda at Money20/20 confirmed that fraud coverage must be a roadmap decision, and the companies growing fastest in payments, fintech and lending need to treat every new product, market and customer segment as a fraud surface that requires coverage from day one. By next year, the question will have moved on from whether risk teams have enough data to effectively operationalize and instead, move toward if they have the ability to deliver on the full picture that sets up their company for success.
Catch Fraud Earlier
Every signup, login and transaction carries evidence that a traditional check never reads. Talk to a SEON expert about what 1,100+ real-time signals would reveal across your customer base.
