Zero-Knowledge Proof
A zero-knowledge proof is a cryptographic protocol that convinces a verifier of a statement’s validity without revealing additional information beyond that statement. The secret evidence used to prove it can remain private.
Also known as: ZKP, zero-knowledge proofs
For example, a person could prove that an issuer-signed credential contains an age above 18 without revealing the exact birth date. The statement must include the credential’s authenticity: proving a claim about an arbitrary number would not establish the person’s age. Public inputs and the fact being proved still reveal information.
In episode 65, Charles Guillemet discusses proving that a policy engine ran its specified computation. In an illustrative agent design, the proof could bind an authorization result to a particular recipient and payment amount while keeping parts of the policy private.
The proof checks the encoded statement under the protocol’s assumptions. It does not establish that the policy captures the user’s intent, that external inputs are truthful, or that a different action was not executed afterward. Those boundaries still need authorization and system checks.
Sources
- Oded Goldreich: Zero-Knowledge tutorial — Defines learning nothing beyond the assertion and discusses proving correctness of secret-based actions.
