Zero-Knowledge Proof (ZKP)

If the statement is true and both the prover and verifier follow the protocol honestly, the verifier will always be convinced of the statement's truth.

• Mechanism: The honest prover possesses a valid witness (secret input) that satisfies the circuit constraints.
• Practical Implication: A legitimate user with correct credentials will never be falsely rejected during a privacy-preserving authentication flow.
• Example: In a ZKP-based login, a user who actually knows the password pre-image will successfully authenticate 100% of the time.

If the statement is false, no cheating prover can convince the honest verifier that it is true, except with some negligible probability.

• Knowledge Soundness: A stronger variant requiring that a prover must actually know the witness, not just stumble upon a convincing argument.
• Computational vs. Statistical: Computational soundness relies on the prover's bounded computational power; statistical soundness holds against unbounded provers.
• Example: A malicious actor cannot forge a proof of solvency for an exchange unless they actually possess the claimed reserves.

The verifier learns absolutely nothing beyond the single bit of information: 'the statement is true.' No other data about the secret witness is leaked.

• Simulator Paradigm: Formally proven by demonstrating the existence of a simulator that can generate a transcript indistinguishable from a real interaction without access to the secret.
• Perfect vs. Computational: Perfect zero-knowledge reveals zero information; computational zero-knowledge reveals an amount computationally infeasible to extract.
• Example: Proving you are over 18 without revealing your exact birthdate, name, or address.

While not part of the original three properties, modern ZK-SNARKs add a critical fourth property: the proof size is small and verification time is fast, regardless of the complexity of the statement being proved.

• Constant Size: Proofs are often just a few hundred bytes, independent of the circuit size.
• Sub-linear Verification: Verification time grows logarithmically or remains constant relative to computation size.
• Practical Impact: Enables on-chain verification of complex off-chain computations on Ethereum layer-2 rollups, where gas costs are directly tied to verification complexity.

Early ZKP systems required extensive back-and-forth interaction between prover and verifier. Modern constructions eliminate this requirement entirely.

• Fiat-Shamir Heuristic: Transforms interactive protocols into non-interactive ones by replacing the verifier's random challenges with the output of a cryptographic hash function.
• Single Message: The prover generates a single proof object that can be verified asynchronously by anyone, at any time.
• Audit Trail Application: An AI system can generate a single non-interactive proof of compliant execution that auditors can verify offline without connecting to the live system.

ZKPs provide a cryptographic guarantee that a specific computation was executed correctly on given inputs, without re-executing it.

• Arithmetic Circuit Representation: The computation is compiled into a set of polynomial constraints over a finite field.
• Verifiable Computation: The verifier checks the proof in milliseconds, even if the original computation took hours.
• Enterprise Use Case: A regulator can verify that a bank's risk model was run correctly on proprietary data without accessing the model weights or the sensitive input data.

A piece of cryptographic evidence that prevents an entity from denying its involvement in a specific action. In the context of ZKPs, a prover can generate a proof that serves as a non-repudiation token—verifiable by anyone, yet revealing nothing about the underlying data.

• Combines digital signatures and trusted timestamps
• Critical for legal accountability in automated AI decisions
• ZKPs enable non-repudiation without data exposure

The process of embedding a cryptographic hash of an audit log into a public blockchain transaction. When combined with ZKPs, organizations can prove compliance to regulators by anchoring a zero-knowledge proof of correct execution on-chain, rather than exposing sensitive operational data.

• Leverages public ledger immutability for independent verification
• ZKP anchoring proves validity without revealing transaction details
• Used in smart contract audit trails for automated governance

A tamper-evident, cryptographically-secured digital credential conforming to W3C standards. ZKPs power selective disclosure in VCs—allowing an AI system to prove it holds a valid compliance certification without revealing the certifying body or expiration date.

• Enables privacy-respecting presentation of audit certifications
• BBS+ signatures provide efficient selective disclosure
• Foundational for decentralized identity in AI governance

An append-only, publicly auditable ledger that records cryptographic commitments to data. Projects like Sigstore use transparency logs for keyless signing of software artifacts. ZKPs extend this model by allowing auditors to verify log consistency and correct operation without accessing the full log contents.

• Enables monitoring and verification of log integrity
• ZKPs provide privacy-preserving consistency proofs
• Used in Certificate Transparency and software supply chain security

A cryptographic fingerprint generated from the inputs, outputs, and version of an AI model during inference. When paired with a ZKP, an organization can prove that a specific model version produced a given output for a given input—without revealing the input data itself.

• Creates a verifiable and non-repudiable prediction record
• ZKPs enable privacy-preserving inference verification
• Essential for right to explanation compliance under GDPR

A hardware-based security paradigm that protects data in use by performing computation within a Trusted Execution Environment (TEE). While TEEs provide confidentiality during processing, ZKPs provide verifiability of the computation's correctness—together forming a comprehensive privacy and integrity framework for AI audit trails.

• Hardware Security Modules (HSMs) protect signing keys
• ZKPs add mathematical verifiability to hardware-based trust
• Combined approach addresses both confidentiality and integrity