Zero-Knowledge KYC: A Short History of Collect-and-Retain, and the Case for Proof of Clean Hands

TL;DR

• Privacy and compliance were never a real trade-off. The identity store that forces the choice is an accident of 2001-era technology, not a requirement of the law.
• Collect-and-retain KYC turned that store into a standing liability: breached on an industrial cycle, weaponizable by AI, and exposed to harvest-now-decrypt-later quantum attacks.
• Proof of Clean Hands proves you are a real, unique, non-sanctioned person without revealing your identity to the parties you transact with. The underlying data stays encrypted, never left readable in the clear.
• Accountability is built into the same proof: a person stays recoverable, but only when a public, pre-committed, onchain condition is met, one record at a time, with every disclosure logged. A lawful access window.
• The ask to regulators is not deregulation. Keep the obligation, drop the identity store.

Proof of Clean Hands and the spirit of the law made executable.

In 2015, twenty-one and a half million Americans were told their lives had been stolen. To hold a federal job, each had filled out a background-check file: every address, every relative, every debt, every past mistake, and for more than five and a half million of them, their fingerprints. They had handed over the most intimate record of who they were, to prove they could be trusted. A foreign intelligence service took all of it from a single government database, in one of the largest breaches of personal data in US history. The breach was not the failure. The collection was: the data sat in one place, in readable form, because that was the only way anyone knew to verify a person and keep a record.

Privacy and accountability are not opposites. We have been told they are for twenty-five years: prove you are not a criminal by surrendering your identity, or keep your identity and accept that no one can be held to account. Pick one. It is a false choice, and we have paid for it in breached databases ever since.

The trade was never a law of nature. It was an accident of timing. In 2001, the only practical way to enable a lawful investigation after the fact was to store the underlying document in the clear. So every regulated wallet, exchange, and bank built the same thing, a store of identities sitting readable and waiting to leak. The breach cycle that followed was not the price of the obligation. It was a side effect of how we met it. The identity store was incidental. And what is incidental can be removed.

Removing it is now a matter of construction, not aspiration.

There is an older principle worth returning to. The state can act against you when you break the shared rules, and only then, through due process aimed at a specific person. It does not get to watch everyone, all the time, on the chance that someone, somewhere, might offend. That is what the Fourth Amendment’s ban on general warrants means in plain terms: no dragnet.

Collect-and-retain KYC inverts it. To participate at all, everyone hands over everything up front. Exposure becomes the default, and privacy is what you have to beg back.

Proof of Clean Hands restores the original architecture, and makes it executable in code.

How we got the identity store

The history is worth a paragraph because the design we live with is an artifact of a specific moment, not of the obligations themselves.

Through the late 1990s, US payments law focused on consumer protection. The Electronic Fund Transfer Act gave you dispute rights. Gramm-Leach-Bliley demanded confidentiality of financial data. E-Sign recognized electronic signatures. Identity, where it appeared, was about protecting the person from the system.

The Patriot Act in 2001 changed the purpose. Identity verification was repurposed from consumer protection to economic-policy enforcement. KYC and AML became mandatory. Monitoring, suspicious-activity reporting, and sanctions screening became fixed costs of operating a regulated wallet. The EU followed with AML directives, then MiCA. The technology stack everywhere defaulted to collect-and-retain.

The obligations were legitimate, and they remain so. The identity store was incidental, an artifact of when and how the rules were built. It is also, now, removable.

Why the 2001 posture is now a liability

Three things have shifted underneath the architecture since 2001. Each one raises the cost of holding identity data in the clear, and none of them was in view when the rules were written.

Breach has gone industrial

The aggregate of mandated identity data, across every regulated entity, is one of the largest concentrations of sensitive personal information in the economy. The data collected to protect the system has become one of its largest attack surfaces. The scale is not hypothetical. In 2024, US organizations reported more than three thousand data compromises and over 1.7 billion victim notices, at an average cost above ten million dollars per US breach. Identity verification is now its own breach category: in 2025, a single misconfigured KYC vendor left roughly one billion identity records exposed across twenty-six countries, a KYC data breach at national scale.

AI weaponizes the data

More than half of all internet traffic is now automated, and stolen credentials are the way into more than a fifth of all breaches. They enable synthetic identities, precision phishing, and re-identification at a scale that was infeasible when the rules were written. On top of the breach risk they have always carried, identity data stores are now training data for the next generation of attack.

The long shadow of quantum

Not all encryption is quantum-safe. Most of what protects data today, RSA and elliptic-curve cryptography, rests on math a large enough quantum computer could eventually break, and the post-quantum schemes that resist it are barely deployed yet. That is why adversaries store encrypted data today on the assumption they will decrypt it later.

Mosca’s inequality gives the rule of thumb: add the confidentiality lifetime of your data to the time you need to migrate to post-quantum encryption. If that sum exceeds the time until quantum arrives, the exposure is already here. Plaintext stores are forfeit on day one. Encrypted stores with multi-decade confidentiality lifetimes are exposed too, including some that our own tools keep for lawful recovery: swapping plaintext for ciphertext removes the day-one exposure, but those records still owe a migration to post-quantum encryption.

The collect-and-retain architecture has plausibly crossed from net-protective to net-liability. The systemic and supervisory risk concentrated in these identity stores may now exceed their investigative value. Keeping the obligation while shedding the identity store is, by now, ordinary risk management.

What Proof of Clean Hands does

The alternative is simple to state: keep the obligation, drop the identity store. Instead of collecting a person's identity and holding it readable, you let them prove they are clean, once, and keep the underlying data encrypted. That is zero-knowledge KYC, and Proof of Clean Hands is how it is done.

Proof of Clean Hands (PoCH) is human.tech's zero-knowledge KYC and AML (ZK KYC): it verifies clean, non-sanctioned participants — real, unique, over 18, eligible — without handing over or storing the documents that show it. It is built on Human ID, the identity protocol previously known as Zeronym, now part of Human Passport.

A person verifies once that they hold a valid government credential and are a unique living human, and is screened against international sanctions, politically-exposed-persons (PEP), and watchlist sources. That check runs through a centralized KYC processor: the underlying data leaves the device and is seen there.

This is where it breaks from collect-and-retain. No plaintext identity data is left in the clear. It is encrypted at issuance, and the credential the user carries is a set of hashed fields, with no raw document readable onchain.

From then on, the party you transact with learns nothing about who you are. You prove only the specific statement a given context needs: that you are a real, unique, non-sanctioned human, over 18, not a US person, in an eligible country. The verifier gets a revocable, reusable attestation, a cryptographic 'yes', never the underlying data, no name, no date of birth, no document image, no face to retain, leak, or subpoena. For a compliance team, that is data minimization by construction, the direction data-protection regimes from GDPR to CCPA have been pushing toward for years.

Accountability is built into the same proof. Alongside the claim that a person is clean, it carries a cryptographic commitment: their identity is encrypted to a threshold network of independent operators, recoverable only when a pre-committed condition, recorded transparently in an onchain contract, is met. That recovery is a rule the user agrees to up front and the code enforces, not a matter of trust or discretion.

The three properties, read together

They are:

• Consent with agency: the user agrees to the recovery condition up front, so nothing opens on an operator's discretion.
• Privacy as the default: you prove only the specific claim a context needs, and reveal nothing else.
• Transparent accountability: recovery runs only under a public, pre-committed, onchain condition, with every disclosure logged.

Privacy and recoverability are established in the same step, so neither can be removed without breaking the other. That is what makes Proof of Clean Hands one cryptographic object rather than a privacy tool with a side-channel for the authorities. It is also what sets it apart from the alternatives: mixers give privacy with no accountability, view-key models rest the privacy on whoever holds the key, and association sets prove something about the funds but never the person. Proof of Clean Hands binds a recoverable identity to privacy in that single object, opened only under a pre-committed, onchain condition.

These are the spirit of the law made executable: the state does not get to watch the law-abiding, but it keeps a narrow, auditable path to the specific person who breaks the contract, one that opens only when a pre-committed onchain condition is met, subject to lawful process.

We mean this as an analogy, not a doctrinal claim. The Fourth Amendment binds the state, and a conditions contract is a private access policy, not a neutral magistrate. The aspiration is to move from "trust an institution not to dragnet you" toward "the system opens one record at a time, only when a public, pre-committed condition is satisfied, subject to lawful process."

That is a shift in the architecture of trust: the institution is no longer the bottleneck your identity must pass through to be safe. The cryptography is.

Learn more about how Proof of Clean Hands works in the deep-dive introduction.

---

Frequently asked questions

What is zero-knowledge KYC?

Zero-knowledge KYC (ZK KYC) verifies that a person meets a requirement — real, unique, over 18, not sanctioned — without revealing or retaining their identity documents. Instead of collecting and storing a copy of who you are, the verifier receives a cryptographic proof. Proof of Clean Hands is a zero-knowledge KYC: the check runs once, the data is encrypted at issuance, and the party you transact with learns only the claim they need.

Is there an alternative to traditional KYC?

Yes. Proof of Clean Hands is a zero-knowledge KYC alternative: it meets the same identity and sanctions obligations without collecting or retaining your documents. You prove you are a real, unique, non-sanctioned person, the underlying data stays encrypted, and disclosure happens only under a public, pre-committed onchain condition. Same obligation, no readable store to breach.

Why do KYC data breaches keep happening?

Because collect-and-retain KYC requires holding a readable copy of every customer's identity documents, and that store is a high-value target. A KYC data breach exposes not just an account but the underlying identity — name, document, biometric — reusable for fraud. ZK KYC removes the store, so there is no readable copy to breach.

Is Proof of Clean Hands a backdoor?

No. A backdoor is secret, single-party, and ungated. Proof of Clean Hands is built to be the opposite on all three: the conditions that permit a decryption live in a public, onchain contract, decryption is per record and never a bulk capability, and every disclosure is logged. Today one authorized party, Holonym Foundation, can allow a decryption, and only when those onchain conditions are met, one record at a time, under a strict daily limit. The data is encrypted to Human Network's independent nodes and held in shares, so no node ever sees the plaintext and the authorizer cannot read a record without the network. What is on the roadmap is decentralizing the authorization itself: an independent forensics DAO of forensics, security, and legal firms approving each request by multisig, so no single party can allow a decryption. The full objection, what it does not defend against, and where the architecture is still unfinished, are in a companion piece, Is Proof of Clean Hands a backdoor? (coming soon).

How is Proof of Clean Hands different from traditional KYC?

Traditional KYC collects and retains your documents in a database, a store of identities that can later be breached. Proof of Clean Hands proves you meet the same rules without leaving your identity data readable: it is encrypted at issuance, and can be disclosed only when a public, onchain conditions contract authorizes a specific request. Disclosure is not automatic even then: someone has to authorize the request, and every disclosure is logged. The data is encrypted to Human Network's independent nodes and held in shares, so no node sees the plaintext. Today the authorizer is Holonym Foundation; on the roadmap that authorization moves to an independent forensics DAO (forensics, security, and legal firms), so no single party can allow a decryption alone. It is privacy-preserving accountability, not surveillance with extra steps.

What does Proof of Clean Hands check?

That a person holds a valid government credential, is a unique living human, and is not on international sanctions, politically-exposed-persons (PEP), or watchlist sources. The verifying institution receives a cryptographic attestation, never the underlying identity data.

Is Proof of Clean Hands live, open source, and audited?

Yes. Clean Hands credentials are in production on Optimism mainnet and Sui, and the code is open source. A privacy-preserving bridge that puts Clean Hands to work has been independently audited by Nethermind Security (report NM-0756, May 2026), and it is now live on Aztec mainnet.

Who can decrypt a verified identity, and when?

Three roles keep it honest, so no one party acts alone. A requester, for example law enforcement with a valid court order, asks. An authorizer validates that request against a pre-committed, public, onchain conditions contract, one record at a time, with every disclosure logged, and is the only party that ever sees the plaintext. The data itself is encrypted to Human Network's independent nodes and held in shares, so no node ever sees the plaintext and the authorizer cannot recover a record without the network. Today the authorizer is a single party, Holonym Foundation, under a strict daily limit. Decentralizing that authorization into an independent forensics DAO of forensics, security, and legal firms approving by multisig — so no single party can allow a decryption — is on the roadmap, not yet in production.

---

human.tech is a suite of technologies designed to enhance personal freedom, privacy, and financial autonomy. Proof of Clean Hands is open source and live today; the documentation and further reading are linked below.

Further reading: Proof of Clean Hands: the deep dive · The Human Network, explained · Proof of Clean Hands documentation · Coin Center, Tear Down This Walled Garden · NIST SP 800-63-4