What is Project Aikya?
Aikya is a loan fraud detection initiative from Sahamati Labs, the innovation arm of Sahamati — the RBI-recognised Self-Regulatory Organization (SRO) that governs India's Account Aggregator (AA) ecosystem. The Account Aggregator framework lets individuals consent to sharing their financial data across institutions through a standardized, consent-driven pipeline rather than ad hoc document uploads or screen-scraping.
Aikya's stated purpose is narrower and very concrete: it helps lenders detect loan stacking — borrowers taking out multiple loans across different lenders in a short window, often to mask over-leverage or commit fraud — along with related fraud patterns.
The problem: loan stacking is a cross-lender problem
Loan stacking is structurally hard for any single lender to catch. Each lender only sees its own book; a borrower who is quietly accumulating debt across five different NBFCs looks fine to each of them individually. Catching the pattern requires visibility across lenders — but no lender wants to hand its loan book, or its customers' financial data, to a competitor or a third-party aggregator wholesale. That tension between "we need to collaborate to see the fraud" and "we can't centralize sensitive customer financial data to do it" is exactly the shape of problem privacy-enhancing technologies (PETs) exist to solve.
Why Trusted Execution Environments
Sahamati Labs' public materials describe a broader PET strategy for the AA ecosystem built on three pillars: Confidential Computing (Trusted Execution Environments, or TEEs), Homomorphic Encryption, and Secure Multi-Party Computation (SMPC). Aikya is positioned as an application of that strategy to loan fraud detection specifically, with TEEs as the enabling primitive.
The appeal of a TEE for this use case is straightforward: a TEE is a hardware-isolated environment where code can process data with a cryptographic guarantee that not even the infrastructure operator can see inside it while it runs. That lets a cross-lender fraud check happen on hardware neither lender individually controls, without either lender's raw loan data ever landing on a foreign compute environment in the clear. Each participant can cross-reference stacking risk against pooled signals while retaining a defensible, auditable claim that no other party read their raw data.
Where this fits in the Account Aggregator ecosystem
The AA framework was built to standardize consented financial data sharing between a customer's institutions and requesting entities. Aikya extends that same ecosystem toward a use case the AA flow alone doesn't naturally solve: fraud signals that only become visible in aggregate, across many lenders, rather than in any single consented data pull. Sahamati Labs frames this as part of "advancing the Open Finance ecosystem in India," convening financial institutions, technology providers, and policymakers around real-world PET deployments rather than research demos.
What we don't yet know
To be transparent: Sahamati's public materials on Aikya are still light on implementation detail — the specific TEE technology in use, how models are trained or updated, what signals are shared into the enclave versus kept local, and any published accuracy or fraud-catch-rate figures are not yet public. This piece is not a PryvX engagement or an endorsement of specific technical claims; it's a note on a real-world example worth watching, because it's one of the clearest illustrations in Indian fintech today of TEEs being applied to a genuine cross-institutional data problem rather than a single-tenant one.
Why it matters
Loan stacking detection is a useful stress test for any PET-based collaboration pattern: the data is sensitive, the participants are commercial competitors, the regulator cares, and the fraud signal genuinely doesn't exist without cross-party visibility. As more of India's financial infrastructure builds on the Account Aggregator rails, initiatives like Aikya are a signal that confidential computing is moving from an infrastructure talking point to a production requirement for exactly the kind of multi-party financial collaboration PryvX builds for.
