Real-Time Payments = Real-Time Fraud
- The growth of instant payments is straining fraud prevention and detection
- The Federal Reserve has begun its launch of FedNow
- Efficient data acquisition and processing will be key to controlling fraud
As we've discussed in previous posts, consumers now have the expectation of near-instantaneous payments solutions. With the popularity of payments like Paypal and, most recently, apps like Venmo, Zelle, and CashApp, consumers are able to pay for items they need/desire and move money at a touch of a button.
For financial institutions, real-time payments have seen slow adoption. However, the Federal Reserve is looking to change that with the recently confirmed FedNow -- the government’s version of real-time payments (RTP) -- which will be deployed in phases, with the first component already available.
While the idea of instantaneous payments is great, there are major fraud implications.
Preventing Fraud While Meeting the Real-Time Payments Challenge
Along with the instant-payment expectation comes, of course, a much more difficult fraud-prevention landscape. Fiserv explores these new challenges:
In the early days of ACH, an institution had up to three days to review transactions before completion. But with the advent of next-day payments, and now real-time person-to-person (P2P) payments services such as Zelle®, fraud detection and mitigation should take place in seconds.
How can financial institutions do this more effectively? The question is especially pertinent because fraud attacks are becoming more sophisticated.
We've seen that the government is not impervious to fraud. According to NBC News, "More than $200 billion in COVID relief loans and grants were distributed to potentially fraudulent actors, nearly one-fifth of all Small Business Administration funds disbursed in the U.S., according to a new estimate by the inspector general for the SBA."
With all this in mind, has the Federal Reserve taken the necessary steps to prevent fraud in the new FedNow system?
Real-time Fraud Detection: Data and Risk Scoring
Whether it's money-moving apps or real-time payment rails, Fiserv cites data management as a key tool in controlling criminal activity:
There is no silver bullet for stopping fraud, but a risk strategy that considers potential vulnerabilities and specific mitigation actions may help reduce exposure. Layered security across different risk events should be a part of that risk strategy. The solutions your financial institution uses for real-time payment fraud mitigation should have the following characteristics:
Intelligence from consortium data. Billions of card and P2P transactions are processed around the world daily, and they all leave a digital footprint. Your financial institution needs instant access to both internal and external sources of this transaction data to evaluate the integrity of real-time transactions.
Real-time risk scoring and interdiction. Transaction time is short; each payment must be evaluated while the transaction is in progress, so real-time decisioning and risk mitigation actions can take place before the payment is submitted for processing. Of course, that’s why you need the big data access.
Fiserv goes on to point out that, given ever-improving data gathering and evaluation, a host of factors can be incorporated into a risk score calculation associated with a payment transaction, including:
- sender and recipient information
- device reputation and mobile device ownership data
- known discrepant data
- geolocation information
The evaluation process, Fiserv points out, allows for the automated discovery of patterns across large volumes of streaming transactions.
Artificial Intelligence: Real-time Fraud Detection
In order to perform real-time fraud detection -- particularly when it comes to large amounts of data -- financial institutions need to leverage artificial intelligence powered by GPUs.
As transactions flow into the system, fraud systems can leverage APIs to transmit the data to several fraud engines -- behavioral analysis, data analytics, image forensics (for check fraud), etc. -- that analyze the transaction in real-time and return results within seconds. With the proper business rules, these results can be scored within the fraud platforms and transactions identified as fraudulent will be flagged, with approval needed from a fraud analyst.
This process enables transaction analysis in real-time, while also reducing losses associated with fraud.
