A16z Rethinks KYC: Institutions Embrace On-Chain Without Being Confined by Traditional KYC Frameworks

Written by Maggie Hsu and Michele Korver, a16z crypto

Translated by Chopper, Foresight News

The "Know Your Customer" (KYC) check is often viewed as an obstacle to traditional finance accessing on-chain ecosystems and driving the overall development of DeFi.

For those unfamiliar with the terminology, these verifications are a necessary component for financial institutions to fulfill their anti-money laundering (AML) and counter-terrorist financing (CTF) obligations under the Bank Secrecy Act (BSA). This process requires coordination among card issuers, verification providers, and regulatory bodies, each with distinct incentives, timelines, and legal requirements.

For banks, brokerages, and other institutions bound by the BSA, the challenge of bringing tokenized products on-chain lies in how to comply with regulations while reaching more customers and accessing liquidity, as compliance demands can add time and complexity. For example, when new customers sign up, institutions often have to repeat identity verification even if the user has already completed similar checks elsewhere. Fortunately, newly developed KYC tools built specifically for blockchain promise to make verification results reusable, freeing us from legacy traditional finance models.

The starting point of the discussion is clarifying the entities that institutions need to identify. In the traditional finance system, institutions conduct transactions through a network of intermediaries that have already undergone identity verification, with each institution independently fulfilling its own KYC obligations. This logic applies equally to on-chain scenarios: a bank's customer identification obligation only applies to its own clients, not every participant in the underlying settlement infrastructure.

The core of customer identification rules is determining who holds an account at this bank. Under this interpretation, users do not become a bank's clients simply by trading with bank customers or swapping assets with pools providing liquidity for the bank. As long as a bank establishes a suitable compliance framework covering its own clients and related transactions, it can meet its BSA compliance obligations.

The U.S. Treasury is currently exploring how financial institutions can leverage such new technologies and methodologies to meet AML/CTF requirements, including the Customer Identification Program (CIP) and KYC. With the passage of the GENIUS Act, regulators need to study and implement innovative solutions and standards to more efficiently achieve modern AML/CTF objectives.

This article will first analyze the legal conditions and practical prerequisites for implementing reusable KYC, then explore the roles played by tools such as zero-knowledge proofs, verifiable credentials, non-transferable tokens, and on-chain data analysis. Finally, we will look ahead to "Know Your Agent": identifying AI agents, the entities they represent behind the scenes, and the operational permissions granted to them.

Applying KYC to the On-Chain World

Traditional KYC processes are conducted independently by each financial institution. Each platform verifies identity through the CIP during user registration, maintains independent records, and conducts ongoing customer due diligence. Identity information is held by the institution, making it a trusted intermediary. When users open accounts on new platforms, they must repeat the verification process, even if assets are circulating between already-verified individuals and enterprises.

Blockchain transactions add another layer of complexity. Asset transfers across chains, protocols, and wallets often lack intermediaries to execute KYC. A single wallet address could come from an exchange subject to the BSA, a DeFi protocol, or a consumer-facing application. These services each have their own risk frameworks and liability allocations, with differing rules across jurisdictions. Some intermediaries themselves are not subject to KYC obligations. Even where KYC is mandatory, there is currently no unified regulatory framework allowing Institution A's verification results to be directly transferred to Institution B. Institutions are forced to repeatedly verify the same users, increasing costs and delays.

Former SEC Commissioner Stephanie Peirce recently stated, "The approach to combating illicit finance hasn't changed: gather as much data as possible. This method has proven ineffective, technological advancements have far outpaced this outdated system, and now is the time for change."

Financial institutions, blockchain developers, and regulators are building new solutions to meet KYC and AML/CTF goals. One direction is allowing institutions to reuse identity verifications completed by other entities. Implementing this mechanism requires clear rules defining which entities' verification results can be relied upon and who bears compliance responsibility.

Reusability and the Trust Ecosystem

Users hope to undergo KYC verification once and use it everywhere. This model is also referred to as reusable KYC, trust-based KYC, or hosted KYC.

Off-chain scenarios also involve repeated verifications. If an individual opens accounts at two different banks, both will require KYC completion, though average users do not frequently open new bank accounts. However, in crypto, users continuously adopt new products without a trusted intermediary ensuring KYC results remain valid throughout the entire transaction lifecycle.

Institutions do not seek to eliminate KYC; they aim to avoid duplicative KYC. Every time a user joins a new exchange or application, the KYC process starts from scratch. This is costly for platforms and hinders user experience; moreover, as deepfake tools become widespread, the reliability of traditional verification methods continues to decline. Banks and other institutions bound by the BSA are weighed down by this expensive, privacy-invasive verification mechanism that struggles to curb illicit finance.

Repeated verification means users must submit sensitive personal information to more institutions. Even submitting digital copies of IDs exposes users to physical security risks, making them more vulnerable to targeted scams, stalking, identity theft, and other unlawful uses of their information. Mobile identity credentials typically disclose a user's full name, address, date of birth, and photo in one go.

Such centralized data "honey pots" storing personal identity information are highly attractive to financial criminals. If information is intercepted, improperly stored, or retained beyond legal limits by institutions, the consequences can be severe. Coin Center notes that the current system forces individuals to "repeatedly submit sensitive personal information to centralized databases, bearing significant personal risk." To avoid these unnecessary security vulnerabilities, privacy advocates urge regulators to consider alternative digital identity verification methods alongside government-issued ID records.

There is a reference case in traditional financial markets: SWIFT's KYC Registry. Global banks have long used the SWIFT messaging network to transmit cross-border payment instructions. In 2014, SWIFT launched the KYC Registry, extending this infrastructure into compliance: banks only need to upload their documents once and can share them with any counterparty in need. The registry now covers nearly 8,000 institutions across 200+ countries. The mechanism works because business occurs B2B, and counterparties are known.

The crypto industry currently lacks a corresponding centralized registry. If Coinbase has already completed a user's KYC to banking-grade standards, can other platforms directly rely on that verification result? If a user whose KYC was completed by Robinhood wants to access an on-chain credit protocol, can Robinhood's prior compliance verification move with the user?

This is what is known as "whitelist liquidity." Accessing pre-verified user pools essentially represents the distribution of compliance capability, but it is not a legal mandate. Regardless of whether clients trade in whitelisted or permissionless pools, the institution's own compliance obligations remain unchanged. Regarding tokenized securities, pool access controls and setting trading eligibility criteria may themselves trigger regulatory obligations under the SEC's latest innovation exemptions. Asset issuers have proactively partnered with major global exchanges, hoping to leverage their compliance frameworks, but the corresponding business architecture is still being built.

Institutions can outsource KYC operations, but the legal classification depends on the service provider and partnership model. According to the FFIEC BSA/AML Examination Manual, reliance is a specific legal provision applicable only when relying on the verification results of another regulated financial institution. If the partner is a non-qualified entity such as an identity verification vendor or credit bureau, it is considered outsourcing. In this scenario, the institution remains solely responsible for fulfilling all CIP obligations.

Similarly, the Financial Action Task Force (FATF), which sets global AML/CTF standards, stipulates that countries may allow financial institutions to entrust third parties to perform certain customer due diligence work, provided three conditions are met simultaneously:

1. The third party is itself subject to AML/KYC-related regulation and supervision and possesses corresponding compliance measures;
2. The relying institution can promptly retrieve underlying KYC raw data upon request;
3. The relying institution bears full legal responsibility for the final compliance outcome.

These rules cannot be directly applied to the crypto industry in full. First, the requirement to "promptly retrieve raw data" implies that regulators or law enforcement must be able to access user PII at any time. This directly conflicts with the privacy-preserving design of most on-chain KYC solutions.

Furthermore, many crypto platforms do not qualify as "regulated institutions" under U.S. federal AML/CIP reliance rules, even if they are registered MSBs, state-chartered trust companies, and bear their own AML obligations. Therefore, they cannot serve as counterparties in formal reliance relationships.

This is an obstacle that must be resolved as the U.S. Treasury advances more flexible, native crypto customer identification solutions. Decentralized approaches are lower-cost, more efficient, better protect user privacy, and reduce leakage risks; however, law enforcement still requires a trusted source capable of confirming real identities and providing documentary evidence for investigations or court proceedings.

In addition to AML program requirements, financial institutions must also comply with fund transfer rules, widely known as the "Travel Rule". Any financial institution or Virtual Asset Service Provider (VASP)—in the U.S., this includes BSA-regulated institutions processing transfers above a threshold, typically $3,000; thresholds in other jurisdictions are lower—must collect and transmit payer and payee identity information to the next institution in the chain. This process flows smoothly in traditional finance because banks already hold KYC data and possess mature messaging channels like SWIFT. In crypto, however, on-chain transactions and KYC information are separated.

Several mature providers offer Travel Rule compliance solutions tailored for the crypto industry, including the TRUST (Transfer Rule Universal Solutions Technical standard) led by Coinbase, along with comparable offerings from Notabene and Sumsub.

In the United States, Travel Rule messaging is only required between financial institutions. Financial institutions and VASPs are not required to send this type of data to non-financial institutions or non-VASP entities (such as self-custody wallets).

When interacting with non-intermediated blockchain technologies like DeFi or self-custody wallets, financial institutions need not overly worry about Travel Rule compliance. Banks or exchanges may still request user information for deposits/withdrawals involving self-custody wallets; this falls under the institution's risk-based AML/CTF program requirements, but the Travel Rule itself does not mandate it.

Positions outside the U.S. vary, and policies regarding transaction thresholds triggering the rule, as well as whether VASPs must collect data from non-VASP individuals/entities, are still being formulated. Regulators worldwide continue to explore how to apply KYC, AML, and the Travel Rule to on-chain scenarios.

The good news is that the U.S. does not need new dedicated legislation. Existing laws and regulatory frameworks can be adapted through supervisory guidance, rule revisions, or innovation exemptions. Congress has already recognized the liability allocation logic discussed above. The GENIUS Act assigns AML and sanctions obligations to permitted stablecoin issuers—the identifiable entities at the application layer—rather than the permissionless underlying networks facilitating stablecoin circulation. Regulators also acknowledge the value of new technology in compliance. In its April 2026 proposal, FinCEN listed machine learning, generative AI, digital identity, blockchain monitoring analytics, and APIs as innovative solutions to meet regulatory obligations.

By endorsing emerging technologies in its proposal, FinCEN signaled that institutions "experimenting responsibly with innovative technologies in their AML/CTF programs will not face significant regulatory scrutiny or enforcement actions solely because of using such innovations." To facilitate practical implementation, the crypto industry has called on FinCEN and banking regulators to issue formal rules or written guidance clarifying when digital identity can be used to satisfy AML program requirements, including CIP/KYC.

Even with clearer regulatory guidance, not all KYC determinations can be reduced to binary cryptographic proofs. Zero-knowledge proofs can verify dates of birth, nationalities, sanction screening results, and credential holdings; however, more complex AML investigations requiring human judgment still rely on traditional due diligence.

Verifiable credentials can assist or even replace parts of the CIP/KYC process, but will not completely replace risk-based customer due diligence. Current Treasury rules already permit non-documentary verification methods, allowing institutions to combine various verification techniques based on their risk assessments. FinCEN's September 2026 FAQ explicitly recognizes that qualified verifiable digital credentials can be used for bank CIP programs, but it does not endorse all decentralized identity or credential designs.

FinCEN and relevant banking regulators can refine guidance within their existing authority. Options include issuing interpretive guidance, adding FAQs, targeted exemptions, pilot programs (approved post-pilot), credential issuer standards, or amending regulations. This approach allows traditional financial institutions, qualified public or private entities to issue verification credentials under uniform standards, reducing the burden on each bank to repeatedly build identical identity profiles.

Until then, financial institutions can already adopt tools compliant with current rules. Below are feasible implementation approaches.

Viable Solutions

Institutions can leverage a trusted, compliant backend system to enable identity credentials to be reused across various on-chain applications. Trusted service providers verify user identity and generate digitally signed credentials recording the verification results. When users present this credential to other applications, the applications only need to validate the issuer's qualifications to confirm if it meets their own verification standards.

This concept is not new, dating back at least to Vitalik Buterin's 2022 proposal for Soulbound Tokens. Supporting standards include W3C-standard Decentralized Identifiers (DIDs) and Verifiable Credentials, which are not exclusive to the crypto industry.

In practice, "certified" compliant KYC vendors, banks, or even governments act as trusted issuers, verifying user identity and issuing credentials. Credentials are stored in user wallets, allowing selective disclosure to needed applications.

Similarly, Succinct Zero-Knowledge Proof systems allow users to prove they have completed KYC without disclosing PII, preserving privacy. Users can reuse this credential provided it meets the receiving institution's CIP requirements, reducing redundant verification.

Verification mechanisms must also be composable, enabling applications to embed verification logic into their own compliance checks. Specifically, to maximize effectiveness, issuers and platforms can embed this logic into token contracts or application-layer smart contracts. Institutions can enforce their own compliance requirements without external components or pushing compliance logic down to the base public chain network.

Implementation Methods

Other viable tools help financial institutions meet risk-based AML program requirements: Know Your Wallet (KYW), Know Your Asset (KYA), and Know Your Transaction (KYT). Exchanges and money transmitter services have already integrated these measures into their broader AML frameworks.

Know Your Wallet (KYW)

KYW binds identity verification to the blockchain addresses used by clients, helping institutions confirm the actual controller of the wallet and which assets/services the client is eligible to access. KYW includes:

• Identity Binding: Typically during account opening KYC, linking real-world identity to wallet addresses;
• Liveness and Uniqueness Verification: Confirming the identity belongs to a real human with no duplicate identities across wallets; periodically refreshing binding relationships to ensure the same real-world identity continuously controls the wallet;
• Access/Permission Control: After wallet verification, obtaining specific access privileges (e.g., accredited investor status, regional compliance eligibility, ability to hold certain asset types), forming the compliance access layer.

KYC in crypto largely follows traditional models. Service providers like Onfido, Veriff, Trulioo, Socure, Incode, ID.me, Persona, Sumsub, and Sardine offer identity binding, document verification, liveness detection, and AML screening. Most of these verifications occur off-chain; verification alone cannot determine which wallet address a user controls, so binding verified identity to a wallet address is an additional step.

On-chain applications can use non-transferable tokens to record verification results, employ zero-knowledge proofs to limit PII disclosure, and use proof-of-humanity schemes to verify users as distinct real humans. Multiple solutions can be combined.

This approach utilizes non-transferable, unsellable tokens, commonly known as Soulbound Tokens, permanently binding them to specific wallet addresses. Within the KYW framework, compliance service providers complete off-chain identity verification and mint a token to the user's wallet as a verification record. Quadrata exemplifies this solution, generating Soulbound Tokens to validate attributes like user nationality and age.

While convenient for auditing verification records, non-transferability alone cannot prevent others from taking over a wallet. For example, if a wallet's private key is stolen and control shifts, the identity tied to the token no longer matches the current controller. Systems also require credential expiration and revocation mechanisms, depending on product implementation; Quadrata supports issuer-driven credential revocation.

Related solutions embed compliance logic directly into asset tokens. DTCC's ComposerX, an institutional tokenization platform, employs the Compliance-Aware Token Framework (CATF), enforcing regulatory and trading rules at execution. Successful asset transfers depend on satisfying compliance conditions.

Some startup projects have planned identity token issuance, such as Burrata. While technically feasible, it suffered from insufficient product-market fit and has since shut down.

Zero-knowledge proofs address another pain point: users can prove identity-related claims without leaking underlying raw data. Numerous solutions exist that generate zero-knowledge proofs via document scanning.

• Self.xyz offers a privacy-preserving "pre-KYC" solution: users scan physical passports to generate zero-knowledge proofs attesting to age 18+ and absence from sanctions lists. It serves as an access screening layer before formal KYC, filtering bots and sanctioned wallets to save time.
• Ethereum privacy L2 Aztec acquired zkPassport, employing a similar approach to generate zero-knowledge proofs from passport scans.
• Opacity Labs utilizes the Zero-Knowledge Transport Layer Security (zkTLS) protocol; its Trusted Data Network (VDN) generates verifiable proofs for any Web2 facts (e.g., account ownership, transaction history), transmitting them on-chain without exposing underlying raw data.

Finally, there are proof-of-humanity solutions. These do not verify who the user is, but only confirm that the wallet is controlled by a unique, real human.

Worldcoin's World ID captures iris biometrics via specialized hardware called Orb to prove humanity uniqueness. Orb encrypts and transmits iris images to the user's phone, deleting the original on-device; it generates hashes compared against others to complete uniqueness verification. Users generate key pairs in the World App, producing zero-knowledge proofs when proving unique human identity.

Pairing proof-of-humanity with identity binding (proving who this real human is) strongly resists deepfakes and reduces identity impersonation in certain scenarios, such as bank account openings. World partnered with Zoom to compare live video feeds with World ID, stamping Zoom accounts with a "Verified Human" badge. Another solution, Human Passport, initially designed to protect Gitcoin grant projects, provides modular identity verification and sybil attack defense tools.

Note: Proof-of-humanity only addresses sybil attacks and deepfakes; it cannot satisfy regulatory identity identification requirements. A complete solution still requires combining other KYC methods.

The above solutions focus on the humans behind wallets. Assets themselves introduce another set of compliance issues, including asset origin, holding restrictions, and transfer permissions.

Know Your Asset (KYA)

KYW focuses on who holds the asset, whereas KYA focuses on what this asset can do and who it can trade with.

Broadly speaking, AML laws are vague. The BSA does not care about asset types; asset classifications are primarily evaluated by market regulators to determine if they are securities, commodities, etc. However, KYA is a vital tool for financial institutions assessing multiple compliance risks. KYA focuses on asset provenance: how the asset was issued, by whom, encompassing compliance attribute verification throughout the issuance phase and its entire lifecycle.

The most direct implementation writes transfer restrictions into the token contract at issuance. RWA tokenization platform Securitize embeds compliance logic at the token level. When issuing BlackRock's BUIDL and Apollo's ACRED, their token contracts mandate transfers only between whitelisted wallets. Assets do not rely on external access layers; the token contract directly enforces transfer restrictions. Similar models are used by Ondo's OUSG and Superstate's USTB: each transaction uses transfer hooks to validate on-chain whitelists, executing settlement only when both sender and receiver pass verification.

An unresolved challenge in KYA is asset provenance: confirming token attributes is not enough; tracing the asset's source is required. Tokens issued by regulated platforms with clean KYA records possess clear origin histories. However, bridging, wrapping, or using tokens as collateral in DeFi protocols damages the provenance chain. While issuance generates clear initial records, maintaining and interpreting these records as assets flow onward remains an unsolved problem, closely related to the KYT transaction due diligence discussed below.

This is more of an operational challenge than a liability trap. The U.S. Office of Foreign Assets Control (OFAC) does not require institutions to assume sanctions risk for every address an asset passes through. Blockchain analytics tools distinguish direct risk exposure (assets moving directly from sanctioned addresses) from multi-hop indirect exposure. Institutions manage the latter through risk thresholds, asset segregation, and Suspicious Activity Report (SAR) filing, rather than rejecting all assets flowing through permissionless infrastructure.

Know Your Transaction (KYT)

KYT helps institutions assess the aforementioned risks, trace connections between different wallets, and map asset movement paths across wallets.

Transaction monitoring is a foundational tool for financial institutions to observe client behavior and flag suspicious activities requiring reporting. For a specific asset, KYT can map its entire lifecycle transaction history, showing which wallets and protocols it has interacted with. Simply put, KYT focuses on reviewing the transactions themselves, operating independently or in conjunction with counterparty identity reviews. This tool satisfies Continuous Customer Due Diligence (CDD) and Enhanced Due Diligence (EDD) requirements.

On-chain analytics providers like Chainalysis, Elliptic, and TRM track wallet linkages and trace fund flows across chains and protocols. Combined with on/off-chain data and emerging AI tools, these products help financial institutions meet BSA and sanctions-related compliance obligations.

Relevant AI tool examples include Eunice.ai, leveraging asset classification frameworks for risk monitoring and disclosure; Didit, offering screening and transaction monitoring with real-time wallet-level risk analysis; and Sardine.ai, conducting initial screening alongside daily continuous transaction monitoring.

Beyond KYW, KYA, and KYT, some blockchains and protocols feature built-in compliance controls.

Blockchain-Level Solutions

Permissioned chains and permissionless public chains are increasingly providing tools for access control and privacy protection. These solutions can intercept transactions before conditions are met while safeguarding sensitive information from leaks, maintaining regulatory audit channels.

The BSA and sanctions regulations do not require validators to be permissioned, the entire network participants to be identified, or blanket access controls enforced across the network. Regardless of the network used, institutions only need to take responsibility for their own clients and transactions. Based on business and product needs, institutions can select from the following solutions, provided they meet compliance obligations.

For example, Canton is a permissioned public network. Under its transaction privacy model, transaction details are only accessible to parties with a legitimate need-to-know, deeply integrating compliance logic with smart contracts. Another example is Circle's Arc, where validators operate under a permissioned model, but the chain itself is public; validators are all real-name participants who have completed KYC/KYB verification. In contrast, permissionless protocols select validators for each transaction internally. Processing a transaction does not make validators counterparties to the institution or client.

Some permissionless protocols also support whitelisted pools, gateways, or rules, easing access control for institutional users. Solana implements compliance via token extensions, using transfer hooks to validate on-chain whitelists for every transfer; trades only execute when both sender and receiver are whitelisted. Enforcement happens at the asset level while the base network remains permissionless, with whitelists typically generated off-chain.

Uniswap v4, as a permissionless protocol, features built-in hooks that attach custom logic at critical points in the swap lifecycle, enabling permissioned pools. The tokens themselves remain permissionless, with only pools imposing access thresholds. Third-party deployment of restricted pools does not convert the protocol into a permissioned chain; the entity deploying the pool and setting access rules bears the associated obligations. Hyperliquid also supports the HIP-3 proposal, allowing deployers to restrict trades to approved addresses.

Decentralized lending protocol Morpho allows users and market participants to build their own compliance controls. The protocol itself does not execute KYC or identity verification. Within Morpho Blue's variable-rate credit agreement, if a market uses whitelisted tokens like KYC-compliant USDC as collateral, compliance enforcement occurs at the asset level, restricting interaction with that specific market to verified users only.

Morpho Midnight's intent-based fixed-rate credit protocol and Morpho Vaults V2 allow more flexible access permissions. Morpho Vaults V2 permits independent managers to set custom thresholds, restricting lender qualifications, enabling capital allocation across networks to borrowers with specific compliance credentials. This modular design allows institutions aiming to trade exclusively with verified counterparties to utilize permissioned markets and vaults without compromising the liquidity efficiency of the broader protocol ecosystem. This is a product and risk preference choice; an institution's own compliance obligations do not depend on whether other participants in shared markets have completed verification.

Cross-Chain Compliance Information Transfer

Access controls determine who can participate in specific markets or vaults. During cross-chain asset transfers, issuers and applications face another challenge: transmitting relevant compliance information to enforce uniform rules across multiple networks.

LayerZero enables cross-chain transmission of "compliance states," eliminating the need to separately rebuild and maintain compliance logic on each chain. Its Omni-Fungible Token (OFT) standard supports custom logic hooks, requiring source and destination chains to complete compliance validation before asset transfer settlement.

LayerZero V2 introduces the Configurable Decentralized Verification Network (DVN) to validate cross-chain messages. DVNs can be configured to enforce compliance checks: verifying if users/wallets completed KYC, screening addresses against sanctions lists, and confirming messages carry Travel Rule metadata before deeming them valid.

Chainlink similarly bridges off-chain verification with on-chain compliance needs via the Chainlink Automated Compliance Engine (ACE), bringing KYC vendor data on-chain. ACE enables reusable KYC, supporting cross-chain wallet binding: credential verification and registration happen once, yet can be invoked multiple times.

Macro Perspective

These tools address different stages of on-chain compliance. Practical implementation requires integration with existing compliance frameworks to help tokenized products reach customers.

The simplest short-term approach is partnering with exchanges that already possess mature compliance systems to reach more customers, though this is not a regulatory mandate. Issuers can also embed control logic directly into the tokens themselves. BUIDL, OUSG, and USDC are prime examples where assets circulate on public chains while remaining governed by on-chain contractual rules.

The next opportunity lies in making compliance verification results reusable within the DeFi ecosystem. Meaning, applications and institutions can rely on verification information issued by other service providers, evaluating them against their own standards. Deploying this system will likely depend on unified protocol standards, compliance-focused software service providers, and traditional platforms with existing KYC users acting as distribution channels.

Future Direction: Know Your Agent (Know Your Agent)

Reusable credentials can verify human client identities. However, as AI agents trade on-chain, manage vaults, and initiate payments on behalf of users, institutions must also verify who authorized the agent and what operational permissions it holds. Know Your Agent extends this discussion to the software executing actions for clients, defining its authority boundaries.

Just as human loan applications require credit scores, AI agent trading requires cryptographically signed credentials. This forms the emerging Know Your Agent framework: KYC for AI agents. Without a robust Know Your Agent framework, liability disputes would arise, creating an awkward situation where either no one takes responsibility or everyone becomes potentially liable.

The Know Your Agent framework requires novel credentials proving agent authorization, represented principals, and operational constraints. Every agent must be identifiable, registerable (binding to individual or corporate entities), possess reputation records, and undergo pre-interaction permission checks. Some solutions run on-chain, others off-chain, but an increasing number share the same cryptographic primitives, including RFC 9421 signatures and W3C verifiable credentials.

On-Chain Solutions

Deployed to the Ethereum mainnet in January 2026, the ERC-8004 standard is the first major on-chain standard for AI agent identity, registering over 30,000 agents in its first week. It provides an on-chain system for AI agent registration, reputation tracking, and recording independent verification results. It mints an ERC-721 token for each agent pointing to its registration file. On-chain records are immutable, trustlessly neutral, and composable, reducing reliance on single platforms.

self.xyz's Self Agent ID builds upon this standard, functioning as an on-chain identity registration system that binds AI agent identity to Self's human identity proofs. Each agent holds an immutable NFT backed by passport verification, which applications can reuse to verify a real human operates the agent.

Advantages of on-chain Know Your Agent solutions: immutability, trust neutrality, and composability.

Off-Chain Solutions

Many solutions use off-chain credentials and signed messages to identify agents and define permissions, even when payments occur on-chain.

• Google's Agent Payment Protocol (AP2) uses signed messages to record user directives, including spending caps and procurement conditions, retaining verifiable evidence of agent authorization.
• Catena Labs' Agent Business Toolkit proposes the ACK v2 scheme, utilizing signed powers of attorney detailing who authorizes the agent and the scope of authority. Every request includes the power of attorney alongside HTTP message signatures (RFC 9421).
• Skyfire KYAPay allows agents to self-prove identity and display represented principals, utilizing signed digital credentials known as JSON Web Tokens (JWT). Skyfire collaborates with Experian's Agent Trust framework to bind agents with verified consumers and devices.
• Circle's Agent Wallet focuses on custody and permission control. Signature key shards remain invisible to the agent, preserving user custody rights. Agents can only autonomously trade within user-set spending limits and recipient restrictions.
• Anchorage's Agent Banking service binds credentials to users or institutions, granting agents spending authority from designated budgets. Transactions are logged under the agent's identity, with permissions revocable at any time.
• Mastercard's Agent Pay registers agents, linking agent payments to consumer authorization and tokenized payment credentials. Partnering with Cloudflare, it uses signed network requests to help merchants verify AI agents accessing websites.
• Visa Smart Commerce replaces card numbers with digital payment tokens, confirming agents hold consumer operational authorization while setting spending limits.
• Zero Hash's Agent Finance Suite screens AI agents and their underlying principals pre-trade. It supports user-delegated permissions and transaction control.
• Ledger is building solutions connecting secure hardware with authorized humans. Its roadmap includes hardware-enforced spending rules, requiring manual approval for sensitive transactions.

Delegating financial operations to AI agents raises questions of authorization and control, returning to the fundamental topic of how institutions apply compliance verification to products and services.

Moving the financial system on-chain also presents an opportunity to optimize compliance frameworks. Most of the above solutions require no new legislation. Existing Treasury rules already accept non-documentary verification methods, and the September 2026 FAQ endorses verifiable digital credentials for CIP. FinCEN has stated that institutions responsibly experimenting with new technologies will not face enforcement penalties merely for doing so. Institutions wishing to avoid redundant KYC processes can now begin implementing existing tools.

Join the official Coincamps community:

X: https://x.com/coincamps

Telegram: https://t.me/coin_camps