Payment Speed Is Outpacing Banks’ Ability To Manage Risk

FedNow processed nearly 5 million instant payments in the second quarter of 2026, an 83% increase from Q1. And its potential reach is expanding.

The Federal Reserve recently issued a proposal that would broaden FedNow’s role in cross-border payments. By allowing participating institutions to use intermediaries outside the Federal Reserve Banks, the proposal would enable a correspondent bank or another provider to handle the international portion of a transaction while FedNow settles the U.S. leg in real time.

For businesses and consumers, that could mean faster access to funds across more international payment flows.

But for banks and fintechs, it raises the stakes of an entrenched infrastructure problem: Payment speeds are advancing faster than the fragmented data systems banks rely on to evaluate risk and defend approval decisions.

Faster rails make fragmented data more consequential

Banks have only seconds to screen for fraud and sanctions exposure before deciding whether to accept a payment through FedNow or another instant payment rail. Cross-border payments make that decision more complex by introducing additional parties, intermediaries, jurisdictions and regulatory requirements without giving banks more time to evaluate them.

Most financial infrastructure isn’t built for that compressed decision window. To evaluate a cross-border payment in real time, automated controls must instantly combine incoming payment information with customer records, transaction histories and compliance data held across internal systems.

Banks’ disconnected legacy systems often prevent automated controls from accessing the full context in time. Fintechs may have more modern infrastructure, but their ability to evaluate payments still depends on data exchanges with banking partners, payment processors and third-party compliance providers that they do not fully control.

And because FedNow settlement is final once completed, a missed signal becomes much harder to address after the funds move.

That puts banks connected directly to instant-payment rails — and fintechs accessing those rails through banking partnerships — in a difficult position: process a payment without the full context or reject a legitimate transaction they can’t verify in time. Neither is a workable foundation for scaling instant payments safely and reliably.

Read More on Fintech : Global Fintech Interview with Rob Young, Managing Director – UK at InDebted

3 priorities for connecting payment data and risk controls

Moving money in real time requires financial institutions to evaluate risk in real time, too. That starts with trusted data, extends to the controls that act on it and ends with a clear record of how each payment decision was made.

1. Trace critical data “back to the piece of paper”

The automated controls used to evaluate real-time payments are only as defensible as the data behind them.

I often put it this way: Can you take the data “back to the piece of paper”? By that I mean a bank should be able to trace critical inputs, such as customer identity information, transaction details and sanctions-screening data, back to their authoritative source.

Following that trail can become complicated as records move across legacy systems. Data may be combined, reformatted, enriched or corrected until it becomes, as I like to say, mishmashed and massaged. Banks should identify the system of record for each input, establish consistent definitions and document how the data changes between its source and the payment control.

This groundwork gives the bank a governed set of inputs it can trust before it begins automating real-time decisions.

2. Build real-time controls around approved data

Once banks have identified and governed the data required for instant-payment risk decisions, they must connect those sources to automated controls that can retrieve and evaluate the approved inputs within seconds.

Testing the end-to-end process under realistic transaction volumes can reveal slow connections, inconsistent formats or weak integrations. It can also confirm whether the broader infrastructure can support instant payments reliably at scale.

AI agents can support this process by examining activity at greater scale, recognizing patterns across transactions and surfacing anomalies that traditional controls might miss. Their role should remain clearly defined: Agents can analyze the data and feed their findings into established risk controls without independently determining whether funds should move.

3. Capture the evidence behind every decision

The final step is to preserve a record of how each payment was evaluated. For every accepted, rejected or escalated payment, banks should automatically capture which data and controls informed the decision, what alerts were generated and why the payment received that outcome.

This creates a defensible audit trail without slowing the payment. If a regulator or investigator questions a transaction, the bank can reconstruct the decision using evidence captured at the time instead of manually piecing together records across multiple systems.

Financial institutions must bring risk decisions up to payment speed

As settlement times shrink, the margin for uncertainty shrinks with them. FedNow can expand the reach of instant payments, but banks and the fintechs building services around these rails still need the data, governance and automated controls to manage risk at the same speed.

The challenge is keeping those decisions defensible when the underlying data may span several organizations. Banks and fintechs need trusted inputs and a clear record of the data, controls and outcomes for which each party is responsible.

Instant settlement changes the timeline for moving money. It shouldn’t change the standard for how decisions are made, documented and defended.

About LatentView Analytics

LatentView Analytics is an AI-driven analytics, data engineering, and consulting company that inspires and transforms businesses to excel in the digital world by harnessing the power of data.