Organizations once had the luxury of reviewing suspicious transactions prior to settlement and clawing them back after the fact. But as both payments and fraud have accelerated, financial institutions are increasingly being pushed to move fraud prevention earlier in the payments lifecycle—ideally before a transaction ever occurs.
In response, U.S. Federal Reserve Financial Services (FRFS)—which operates the FedNow instant payments system—is launching an API aimed at enhancing the security of instant payments. The goal is to provide financial institutions and payments service providers with insights derived from historical FedNow data and network intelligence, helping them determine whether to proceed with a transaction.
Alongside improved fraud detection, FedNow notes that these insights could also enable additional capabilities, such as delivering tailored messaging to users who are on the verge of initiating a high-risk payment.
“This network intelligence API is a step in the right direction,” said Jennifer Pitt, Senior Fraud Analyst at Javelin Strategy & Research. “Network intelligence is key to detecting fraud in real time, and it is key to identifying organized fraud rings. As fraudsters continue to skirt fraud flagging and reporting thresholds by spreading activity across institutions, network intelligence is a critical piece in identifying that behavior.”
The Two Elements
Two factors have allowed criminals to rapidly scale their efforts: technology and organization. Artificial intelligence has played a key role in supercharging fraudulent activity, and the threat is likely to intensify as cybercriminals experiment with frontier AI—cutting-edge models that stretch technological boundaries—to reduce the time, expense, and skill required to run fraud campaigns.
Compounding this issue, these campaigns are often carried out by organized fraud rings, which can amplify their impact. One example is a crypto investment scheme that allegedly defrauded victims of more than €700 million (roughly $817 million).
The Protections at Hand
Unfortunately, these fraud challenges are expected to grow as instant payment systems such as RTP and FedNow continues to surge. Both U.S. instant payment networks have recently reported record highs in transaction volume and value, and FedNow’s model could eventually expand globally.
In many cases, real-time payments are also irrevocable, and don’t offer recourse mechanisms like credit card chargebacks—protections that many consumers have come to rely on. As consumers increasingly expect both immediacy and safeguards, financial institutions are facing mounting pressure to adapt.
Despite these challenges, many financial institutions are still hesitant to fully leverage the fraud prevention tools already available to them.
“Participation in this FRFS network intelligence project is voluntary, that raises a question around adoption,” Pitt said. “Many banks do not fully use existing information-sharing frameworks like Section 314(b) of the USA PATRIOT Act, so it is fair to question whether they will adopt this. Some institutions point to lack of manpower and the fact that 314(b) is not real time.”
“This model may help address those concerns since the information is delivered automatically at the time a payment decision is being made, rather than requiring case-by-case outreach,” she said.
Only Part of the Picture
Another longstanding barrier is financial institutions’ reluctance to share data due to privacy and competitive concerns. The International Monetary Fund has implored banks to reconsider this stance, warning that a fragmented view of fraud is undermining their ability to respond effectively.
This view is echoed by the Global Anti-Scam Alliance, which recently partnered with OpenAI to launch scam.org, a platform offering resources for scam education, reporting, prevention, and victim support. The initiative aims to provide a centralized hub where industry participants can begin building a more standardized response to escalating scams.
Despite growing calls for industry-wide collaboration, progress will require buy-in. It remains to be seen whether tools like the FedNow API will be compelling enough to bring organizations off the sidelines.
“Even though the data is abstracted and network-derived, participation still requires a level of comfort with contributing to and using shared intelligence,” Pitt said. “For some organizations, that hesitation will remain.”
“Another limitation of the FRFS network intelligence mod is that the intelligence is focused on the receiving account,” she said. “That means participating organizations still do not have the full picture of the transaction or the parties involved. What is ultimately needed is a more complete view of both the sender and receiver, including historical and current information about the transaction, device, and account behavior. Without that, organizations are still making decisions with only part of the picture.”
