Project Mandala

Cross-border transactions can require checks against different national restrictions before funds move between institutions. Project Mandala applies compliance by design by placing selected regulatory requirements inside a common transaction protocol. The architecture combines peer-to-peer messaging with a rules engine and a proof engine.

The experimental Mandala architecture checks the applicable conditions before a payment instruction is initiated. After the required checks pass, the proof engine generates evidence that can accompany the instruction or settlement asset. Participating institutions can verify the proof during the transaction process.

Mandala uses zero-knowledge proofs and secure multiparty computation to support checks while limiting disclosure of underlying information. A zero-knowledge proof can demonstrate that a specified condition holds without exposing the private data used to establish that condition. Secure multiparty computation lets participants compute a result from private inputs without disclosing those inputs to each other.

These methods connect zero-knowledge proofs in finance to a specific compliance workflow between institutions. The proof establishes the encoded statement under the chosen cryptographic assumptions. Financial institutions still need reliable source data and an appropriate interpretation of the regulatory requirement being tested.

The first-phase lending example linked a financial institution in Singapore with an institution in Malaysia. The experiment combined sanctions screening with checks against capital flow management measures. Central banks could also examine compliance monitoring through the same experimental framework.

A second example examined financing for capital investment between South Korea and Australia through an unlisted securities transaction. The experiment included sanctions screening and reporting requirements concerning capital flow management. These examples tested particular regulatory processes within defined transaction scenarios, leaving wider institutional compliance responsibilities in place.

Mandala tested connections with a wholesale central bank digital currency environment and traditional payment messaging through Swift. Programmable compliance for digital assets can also be incorporated into smart contracts. This design allows a proof to travel with a transaction while the actual settlement follows the connected payment system.

A wholesale CBDC concerns the central bank money used for settlement between eligible institutions. Mandala examines the compliance checks and evidence associated with a cross-border transaction. The distinction helps teams assess how a compliance protocol might interact with tokenized deposits or existing payment instructions. Project Agorá examines a platform bringing tokenized commercial bank deposits together with tokenized central bank reserves. That settlement-platform research addresses a different layer from the compliance checks examined through Mandala.

Project Mandala continues in Phase 2 under the BIS Innovation Hub Singapore Centre. The participating central banks are the Reserve Bank of Australia, Banque de France, Reserve Bank of India, Central Bank of Kuwait, Bank Negara Malaysia, Bangko Sentral ng Pilipinas, and Monetary Authority of Singapore.

Phase 2 examines benefits empirically, extends the research to broader use cases, and studies usefulness for other platforms and projects. The project remains experimental, with further research into programmable compliance for digital assets. Participation in this research does not establish a generally available payment service or a production deployment across the participating countries.

The Mandala regulatory framework preserves the responsibility of financial institutions to interpret and apply official regulatory measures. Encoding a requirement into software therefore depends on how an institution translates that requirement into a rule. A proof of an outdated rule cannot establish whether the current requirement has been satisfied.

A bank assessing this architecture should examine ownership of rule updates and the evidence supporting each result. Relevant questions include which jurisdiction applies, which source data entered the check, and which rule version generated the proof. These are implementation questions for testing a compliance process, without treating cryptographic verification as a complete legal judgment.