Swift has spent two years showing that banks can learn from each other’s fraud without sharing customer data. Its trials with a group of international banks used federated learning and privacy-enhancing techniques, and reported that detection roughly doubled when models could learn across institutions rather than alone. The next phase, on live data, is on the agenda in Miami. It is good work and the industry should want it to succeed.

It also has a natural boundary. Swift’s network is correspondent cross-border traffic between the banks that sit on it. A great deal of the fraud that reaches ordinary customers never touches that layer. It moves on domestic instant rails, between a Tier-2 bank, a credit union and a fintech wallet, in one currency, in one country. The institutions on that layer are also the ones least able to build a shared-intelligence capability by themselves.

The principle Swift has demonstrated applies there just as well. Data stays where it is. Intelligence moves. Every signal that moves carries its lineage: where it originated, how confident the source is, when it was last confirmed. A receiving institution can weight it accordingly and can explain, later, why it acted on it.

The way this reaches the domestic layer, in our view, is through the aggregators that already serve it. A CUSO, a bankers’ bank or a processor already sits above dozens or hundreds of member institutions. Deploy the scoring layer once at that level, let each member consume it under its own policy, and the network effect Swift is building for cross-border begins to exist for domestic instant payments. A scheme seen at one member becomes a signal at all of them, without any member seeing another’s customers.

Two design choices decide whether this works in practice rather than in a whitepaper.

The first is that shared intelligence has to be an input to the institution’s own decision, never a decision imposed from outside. Liability sits with the institution that holds the customer and the account, so the verdict must sit there too. The aggregator supplies the signal, the reason and the record.

The second is that the identity signal has to be durable. A shared network is only as good as its confidence that the party transacting is the party that enrolled. Continuous identity and fraud risk assessment on every transaction, delivered with a specialist partner where biometric verification and cryptographic credentials are needed, gives the network a common anchor to share against.

We built our platform around these principles when we scored a national real-time rail, and the lineage rule and the institution-decides rule come straight from that experience. Extending shared intelligence to the domestic layer is work we are doing with the aggregators who serve it, and we would rather describe it as a direction than as a finished product. The industry has had enough of the alternative.

About RS Software
As the payments ecosystem transitions toward an AI-driven, real-time, and increasingly autonomous model, where TRUST is the new currency, RS Software (India) Limited's product portfolio modernizes the most critical areas across the value chain, leveraging decades of work building mission-critical payment infrastructure globally. Each product is purpose-built to address specific structural shifts in the market while collectively enabling a unified, AI-led financial ecosystem. RS IntelliEdge™ is designed to address evolving transaction-level controls and shift to continuous, identity-centric intelligence systems, protecting approximately 95% of India’s digital payment transactions and screening billions of transactions each month. RS DigitalEdge™ is designed to enable real-time routing, optimization, and decisioning across payment flows. Payabbhi® is designed to enhance the acceptance and acquiring layer, transforming it into a data-driven revenue engine. www.rssoftware.com