Research note · Research
Selective Disclosure for Agent Finance
Compliance should receive the facts it needs, not the whole context. This note frames disclosure as a policy problem, not an all-or-nothing choice.
Real-world assets bring real obligations. The failure mode of pure privacy projects is treating disclosure as betrayal; the failure mode of pure compliance is handing over everything. Selective disclosure is the policy layer in between: who may learn what, in which scope, for how long, and with what proof that the disclosed fact is true.
Mechanisms range from role-scoped access policies and time-boxed audit keys to verifiable credentials and zero-knowledge proofs that assert a fact ('position exceeded threshold') without revealing the underlying context. Each mechanism shifts where trust sits — in the issuer, the prover or the auditor.
This note stays at the policy level deliberately. It defines disclosure matrices and open questions rather than proposing a protocol, because a protocol proposed before the requirements are honest is marketing.
What this note covers
- Disclosure policies by role, scope and time
- Verifiable credentials and proofs for claims
- Auditor access without full memory access
- Where privacy and regulatory requirements conflict