DeFi Compliance in 2026: How the Rules Get Into the Code

Decentralized finance was built to remove the middlemen that financial rules are written to control. There is no bank to license, no broker to register, no compliance officer to subpoena; there is a set of smart contracts that anyone with a wallet can call. That design is exactly what makes DeFi useful, and exactly what makes it hard to regulate. In 2026 the question has stopped being whether the rules apply and become how they get enforced against software that has no front door.

The backdrop is a market that has cooled. Total value locked across DeFi sits around $70 billion in mid-2026, down roughly 39% since the start of the year as yields compressed and capital rotated out of leveraged strategies, part of the same yield squeeze reshaping validator economics across the sector (CryptoRank). Even shrunken, that is a large pool of money moving without the identity checks that govern an ordinary bank transfer, and regulators on both sides of the Atlantic have spent the year deciding what to do about it.

Three events frame the moment. In July, the Financial Action Task Force published its first dedicated report on DeFi and proposed a test for when a protocol is regulated at all. In Washington, the Senate scheduled a September procedural vote on the CLARITY Act, the market-structure bill that would write a decentralization test into US law. And in New York, prosecutors are preparing to retry Tornado Cash developer Roman Storm on the counts a jury deadlocked over in 2025. Together they circle one question: can code be made to follow the rules without ceasing to be code?

This is a guide to how that is actually being attempted. Not the enforcement map of who gets sued, but the working machinery: the layers of the stack where compliance is being bolted on, the trade-off each one carries, and whether the result is still recognizably DeFi.

What Compliance Means When There Is No Company

Compliance is not one rule. It is five separate regimes, each attached to a different activity and enforced by a different US authority. Conflating them is the most common mistake in DeFi coverage, and it matters because a protocol can be clean under one and exposed under another.

The correction that matters most: anti-money-laundering rules in the United States are enforced by the Financial Crimes Enforcement Network (FinCEN) and the Office of Foreign Assets Control (OFAC), both inside the Treasury, under the Bank Secrecy Act. They are not enforced by the Securities and Exchange Commission. The SEC governs whether a token is a security and whether trading or brokering it needs registration. Those are different questions with different tests, and a headline that says the SEC is cracking down on DeFi money laundering is describing something that does not exist.

For most DeFi protocols the live surfaces are AML and sanctions, because those attach to the act of moving value rather than to a securities classification. That is why the fights of 2025 and 2026 have centered on money transmission and OFAC screening rather than on the Howey test.

The practical consequence is that a single protocol can face several regulators at once, each asking a different question. A lending market might be clean on securities but exposed on money transmission; a perpetuals venue might satisfy the CFTC yet still owe sanctions screening. Compliance in DeFi is therefore less a single license to obtain than a set of overlapping obligations to map, which is why the rest of this guide follows the obligations to the places in the stack where each one can actually be enforced.

The Design Problem: Permissionless by Default

Every existing financial rule assumes a regulated intermediary: an entity that onboards customers, can be licensed, can freeze an account, and can be held to answer. DeFi’s core proposition is the removal of that entity. A permissionless protocol accepts any address without asking who owns it. A non-custodial protocol never takes control of user funds, so it has nothing to freeze. Composability means one protocol’s output is another’s input, so value can hop through a dozen contracts in a single transaction with no human in the loop.

The three classic stages of money laundering (placement, layering, integration) map onto this uncomfortably well. DeFi is weakest at the layering stage, where funds are shuffled to obscure their origin; a chain of swaps, bridges, and liquidity pools can layer value faster and more cheaply than any offshore shell network. The mismatch is structural. You cannot serve a subpoena on a smart contract, revoke its license, or place it on a no-fly list; it has no license and no legs.

Regulators spent the crackdown era trying to find the humans behind the code. The 2026 approach is subtler: find the points in the stack where a person or a company still sits, and make compliance attach there. The rest of this guide is a tour of those points.

The Control Test: How FATF Decided Who Counts

On 21 July 2026 the Financial Action Task Force, the intergovernmental body that sets global anti-money-laundering standards, published a targeted report on the regulatory challenges of decentralized finance. Its central idea is the control-or-sufficient-influence test, quickly shortened to COSI. A DeFi arrangement falls within Recommendation 15, the crypto rule, wherever a natural or legal person exercises control or sufficient influence over it, regardless of the decentralized label (Chainalysis).

The report sorts protocols into three buckets:

• Identifiable controllers: a person or company clearly runs it. That party is a virtual asset service provider (VASP) and owes full AML obligations.
• Effectively centralized, controllers hidden: someone is in control but has not been identified. Still in scope; supervisors are told to go find them.
• Genuinely leaderless: no one exercises control or sufficient influence. Outside the VASP rules, though jurisdictions are still told to monitor residual risk.

What counts as control? FATF lists on-chain signals (governance-token concentration in a few wallets, admin keys that can upgrade contracts, control over fee and treasury flows) and off-chain ones (control of the front-end interface, the development repository, and public statements about the power to change the protocol). The through-line is that control follows the ability to change or profit from the system, not the marketing.

The gap between the standard and reality is enormous. FATF found that around 93% of surveyed jurisdictions had not identified a single qualifying DeFi VASP; only four imposed licensing requirements, and essentially one had taken enforcement action. The test now exists on paper almost everywhere and is enforced almost nowhere. That gap is the whole story of DeFi compliance in 2026: the rules have been written faster than anyone can apply them.

The Tornado Cash Line: Code Versus Conduct

No case has done more to define the boundary than Tornado Cash, the Ethereum mixing protocol. The saga runs on three tracks.

First, sanctions. In August 2022 OFAC added Tornado Cash’s smart-contract addresses to its blocked list, the first time the US sanctioned software rather than a person or entity. Coders and civil-liberties groups sued. In November 2024 the Fifth Circuit ruled in Van Loon v. Treasury that immutable smart contracts do not qualify as property under the statute OFAC relied on, because no one owns or controls them, so they cannot be sanctioned. OFAC formally delisted Tornado Cash in March 2025.

Second, the criminal case. Developer Roman Storm was tried in 2025. In August a Manhattan jury convicted him of conspiracy to run an unlicensed money-transmitting business but deadlocked on the money-laundering and sanctions-evasion counts. In 2026 the Department of Justice moved to retry him on the two hung counts, asking for a slot on 5 or 12 October; together they carry a maximum of 40 years (The Block).

Third, the policy signal. Weeks after the verdict, Matthew Galeotti, then acting head of the DOJ’s Criminal Division, told an industry summit that prosecutors would not bring money-transmission charges against developers of genuinely decentralized software where no third party takes custody of user funds. His summary line, widely reported, was that “merely writing code without ill intent is not a crime”.

Put together, the trilogy draws the working rule of DeFi compliance: publishing code is protected, but operating a service around that code (running the interface, taking a fee, relaying transactions, marketing it) is conduct, and conduct can be prosecuted. The retrial will test exactly where on that spectrum Storm’s activity fell.

Compliance in the Settlement Asset

The most effective place to put a rule turned out not to be the protocol at all. It is the money moving through it.

Most DeFi value settles in dollar stablecoins, a market worth about $287 billion in mid-2026, with Tether’s USDT (around $183 billion) and Circle’s USDC (around $72 billion) making up close to nine-tenths of it (CoinGecko). Those tokens are issued by companies, and issuers can freeze, blacklist, and burn the balance at a specific address. When compliance lives in the asset, it travels with the asset through every protocol the asset touches, no matter how decentralized that protocol is.

The GENIUS Act, signed into law in 2025, made this explicit for payment stablecoins. Issuers must hold the technical capability to comply with a lawful order to “seize, freeze, burn, or prevent the transfer” of the coins they issue, and a FinCEN and OFAC rule proposed in April 2026 extended that expectation to secondary-market activity, including transfers where the issuer is a party only through its smart contract. Foreign issuers serving US persons are covered too. It is one thread in the wider story of how stablecoin rules are landing in 2026, and its practical effect on DeFi is a compliance backstop that does not care whether a lending pool is permissioned: freeze the stablecoin at the wallet, and the wallet’s DeFi positions are stranded.

This is why some builders describe the settlement layer as the real regulatory perimeter. A protocol can be as trustless as its designers like, but if it settles in a token with a kill switch, the state keeps a lever. Privacy advocates see the same fact and call it the frozen account returning by another name.

This is not hypothetical. Both Tether and Circle have frozen balances at specific addresses in response to law-enforcement requests, and both publish the fact that they can. Because a stablecoin is a liability on the issuer’s books, the issuer keeps a permanent switch over every unit in circulation, wherever it sits: in a wallet, in a lending pool, in a liquidity position, or halfway through a multi-step transaction. For a protocol that markets itself as unstoppable, settling in such an asset is a quiet admission that one actor upstream can still pull the plug.

Compliance at the Front-End

If the settlement asset is one lever, the website is the other. A smart contract cannot be censored, but the hosted interface most users rely on to reach it is a company with servers, a domain, and lawyers.

Uniswap Labs, which builds the most-used interface to the Uniswap protocol, screens wallet addresses against sanctions and illicit-activity lists supplied by the analytics firm TRM Labs, blocking flagged addresses (sanctioned, stolen-funds, ransomware) from the front-end while the underlying contracts keep running for anyone who calls them directly. Its token-listing policy also removes assets from the interface on an adverse legal finding, a fraud allegation, or an intellectual-property claim. The contracts are untouched; the shop window is curated.

This is the choke point most of DeFi’s day-to-day compliance actually runs through. After the 2022 Tornado Cash sanctions, several leading front-ends restricted flagged addresses within days, even though the protocols themselves could not be changed. The lesson for builders is blunt: the interface is a regulated surface even when the protocol is not, which is one reason so much 2026 product design pushes users toward self-custody and direct contract interaction, the same shift driving the race to make smart-account wallets native. The relayers and block builders that order transactions face their own version of the question, since whoever chooses which transactions to include, the same actors at the center of the MEV supply chain, can in principle be asked to exclude some.

The front-end is also the layer most exposed to jurisdiction. A hosted interface can be geo-blocked so that users in sanctioned countries never see it, can be served in one country and not another, and can be pulled down by a court order in a way the contracts behind it cannot. That asymmetry is why several protocols now ship more than one front-end: some run by the founding company under full compliance, others published as open-source alternatives that anyone can host, so the protocol survives even if the flagship interface is restricted. It is compliance and censorship-resistance living side by side in the same product.

Permissioned DeFi and the DeFi Mullet

If open protocols are hard to police, one answer is to close them. Permissioned DeFi keeps the on-chain mechanics of a lending pool or exchange but gates access behind identity checks, so every participant is a known, screened counterparty.

Aave, the largest lending protocol, pioneered the model with Aave Arc, which used a whitelister to admit only KYC-verified institutions. Its successor, Aave Horizon, launched in 2025 as a market where real-world-asset collateral is permissioned but stablecoin liquidity stays open, drawing hundreds of millions of dollars in deposits with partners including Circle, Franklin Templeton, and other tokenized-asset issuers. Maple, Clearpool, and Centrifuge run related institutional-credit models. The industry nickname is the DeFi mullet: a compliant, KYC-checked fintech front, with permissionless DeFi plumbing in the back.

The trade-off is explicit. Permissioned pools solve the identity problem by reintroducing the gatekeeper that DeFi removed. They appeal to institutions and to regulators precisely because they look like the regulated system, and they disappoint anyone who wanted DeFi to be open by default. Governance of these pools, and control of the whitelist, becomes its own concentration of power; the same control indicators FATF listed apply, and whoever holds the admin keys holds the compliance obligation. That is why key management, meaning how many signers, and who, has become a governance question in its own right.

The model is spreading because it resolves the institution’s problem cleanly: a bank or asset manager cannot touch a pool where the counterparty might be a sanctioned entity, but it can meet its own rules if every participant is pre-screened. What it does not resolve is the philosophical one. A permissioned pool is DeFi in its mechanics and traditional finance in its access, and reasonable people disagree over whether that is a bridge to the mainstream or a hollowing-out of the original idea. What is not in dispute is that most new institutional capital entering on-chain finance in 2026 arrives through these gated venues rather than open pools.

Privacy Pools: Compliance Without Surveillance

At the opposite pole sits a more ambitious idea: prove your funds are clean without revealing who you are. Privacy Pools, deployed on Ethereum mainnet in March 2025 by the group 0xbow and grounded in a 2023 paper co-authored by Ethereum’s Vitalik Buterin, uses zero-knowledge proofs and association sets to let a user show that a deposit does not come from a known illicit source, without disclosing identity or full transaction history (The Block).

The design tries to thread the needle that Tornado Cash could not. A mixer that pools everyone together, honest and criminal alike, becomes a laundering tool. Privacy Pools lets honest users voluntarily prove membership in a set that excludes flagged deposits, so they gain privacy while illicit funds are pushed out of the anonymity set rather than hidden inside it. Buterin has routed his own transactions through the system to demonstrate the point.

Whether this satisfies the law is unsettled. The open questions are who curates the association sets (a censor by another name, if done badly) and whether a service that facilitates private transfers, however honestly, still counts as money transmission under the reading that convicted Storm. Compliant privacy is a real technical achievement; whether it is a legal safe harbor is a separate question no court has answered.

Programmable KYC and On-Chain Identity

The identity problem has its own emerging toolkit. Instead of every protocol collecting passports, and becoming a honeypot of personal data, the model gaining ground is the reusable, privacy-preserving credential: verify once with a trusted issuer, then carry a cryptographic attestation that proves a fact without revealing the underlying documents.

In practice this means on-chain attestations (a signed claim that an address belongs to a verified person, or that its owner is not in a sanctioned jurisdiction) and zero-knowledge KYC, where a user proves an attribute (over 18, not sanctioned, accredited) without handing over the data behind it. A protocol can then require a valid attestation to interact, satisfying an access rule while holding no personal information itself. The same account-layer programmability that makes this practical is arriving through smart accounts, which can enforce policy at the wallet level rather than trusting the interface to do it.

The appeal is that it decouples compliance from surveillance: the protocol learns that you are allowed in, not who you are. The risk is that reusable identity becomes a universal permission slip that can be revoked, turning a credential issuer into a new gatekeeper with the power to switch off a person’s access across every protocol at once. As with permissioned pools, the compliance moves to whoever controls the attestation.

The Analytics Dragnet

Underneath all of this sits the surveillance layer that makes any of it enforceable: blockchain analytics. Three firms dominate, Chainalysis, Elliptic, and TRM Labs, and their software is what turns a public ledger into a compliance tool, clustering addresses, tracing flows, and scoring wallets for risk. They supply the screening lists that front-ends and exchanges use, and the forensic trails that prosecutors build cases on.

Two caveats matter. First, the scores are probabilistic, not proof; an address flagged as high-risk has been associated with suspicious activity by a heuristic, which is not the same as a finding of guilt, and false positives strand innocent users. Second, the picture the data paints is more modest than the enforcement effort implies. By Chainalysis’s own accounting, illicit addresses received well over $150 billion in 2025, a sharp jump driven by sanctions-related flows, yet that still amounts to less than 1% of all on-chain activity; stablecoins, not Bitcoin, now carry the large majority of that illicit value (Chainalysis).

In practice, screening is a gate applied at the moment of interaction. When a wallet connects to a compliant front-end or an exchange, its address is checked against the provider’s risk lists in real time, and a hit can block the action, freeze a withdrawal, or trigger a manual review. The heuristics behind those lists (which addresses cluster together, which have touched a sanctioned contract, which sit one hop from a known hack) are powerful and opaque in equal measure, and a user rarely learns why a score went against them. That opacity is tolerable for a bank compliance desk and corrosive for a system that promised permissionless access.

That last point fuels the sharpest critique of the whole regime. Peter Van Valkenburgh of the research group Coin Center has argued that the existing anti-money-laundering system “does remarkably little to prevent illicit finance” while imposing heavy privacy and compliance costs, a mismatch he says is only starker in a system where the ledger is already public. Accept that or not, the analytics layer is the quiet foundation the rest of the stack rests on.

Who Is Still on the Hook

For all the engineering, liability in 2026 still follows a simple principle: it attaches to control and profit, not to the label. The precedents map the exposure.

The Ooki DAO case is the sharpest warning. In 2023 the CFTC won a default judgment establishing that a DAO could be treated as an unincorporated association and a legal person under the Commodity Exchange Act, that its governance-token holders could be members, and it imposed a penalty and ordered the group’s website taken offline (CFTC). The message to token-holders who vote is that governance can carry liability, which is one reason serious DAOs now wrap themselves in legal entities (foundations, association structures) rather than resting on the idea that a protocol is nobody.

The reassurance, such as it is, comes from the Galeotti line and the SEC’s shift below: pure development is increasingly protected, and the enforcement energy is moving toward the points of control the compliance stack is being built around.

The Legislative Picture: CLARITY, GENIUS, and the SEC Pivot

US policy in 2026 is trying, unevenly, to replace enforcement-by-prosecution with written rules.

The CLARITY Act, a market-structure bill, would split jurisdiction between the SEC and CFTC and, crucially for DeFi, write a statutory decentralization test into law and define the responsibilities of developers, users, and interfaces. It passed the House in 2025, then stalled in the Senate, where DeFi and illicit-finance provisions are among the sticking points. Senate leaders scheduled a first procedural vote for 15 September 2026; with Republicans holding 53 seats, at least seven Democrats or independents would have to join for it to advance, and prediction markets have been pricing failure (CoinDesk). A yes vote would not make it law; the bill would still need floor passage and reconciliation with the House.

The GENIUS Act, already law, has moved fastest, and as described it routes compliance into DeFi through the stablecoin itself. On the enforcement side, the SEC under Chair Paul Atkins has reversed course from the prior administration, winding down several crypto cases and floating an innovation exemption that would let projects issue and operate tokens for a period without full registration. At a June 2025 roundtable pointedly titled “DeFi and the American Spirit,” Atkins said “the right to have self-custody of one’s private property is a foundational American value that should not disappear when one logs onto the internet”. One more thread closed a compliance surface entirely: the IRS DeFi broker rule, which would have forced front-ends to collect user data and file tax forms, was repealed by Congress in 2025, so interfaces are off the hook for tax reporting even as custodial exchanges still report.

Europe Takes a Different Road

The European Union reached the same destination by a different route, then stopped short of DeFi. The Markets in Crypto-Assets regulation (MiCA) governs crypto-asset service providers and issuers, not autonomous protocols; the European Banking Authority and ESMA judged DeFi a niche of roughly 4% of crypto activity and declined to prioritize bespoke DeFi rules, though a MiCA review consultation opened in 2026 to revisit the question.

Anti-money-laundering sits separately in Europe, in the Anti-Money Laundering Regulation and the Transfer of Funds Regulation, overseen by the new Anti-Money Laundering Authority in Frankfurt, which begins directly supervising the highest-risk crypto firms and, from 2027, bans anonymous crypto accounts and privacy services at regulated providers. For European readers the regulator names change (BaFin, the AMF, Consob, and the rest under MiCA), but the structural point matches the US: the rules attach to the service provider, and autonomous DeFi sits in a gap no one has yet decided how to close.

The detail that will decide Europe’s DeFi treatment is the definition itself. MiCA carves out crypto-asset services provided in a fully decentralized manner without an intermediary, which sounds like an exemption until you ask who decides what fully decentralized means. The same control question FATF poses returns in European clothing: a protocol with a foundation, a fee switch, and an admin key looks a lot like a service provider, and supervisors have signaled they will look past the label. Until the MiCA review lands, builders serving European users work on the assumption that a front-end or a legal wrapper can pull them into scope.

A Practical Map for Builders, Interfaces, and Users

Strip away the doctrine and a few working rules emerge.

• For builders: the safest posture is genuine decentralization plus distance from the money. Renounce admin keys or make upgrades transparently governed; do not take custody; do not take a discretionary fee you control; and remember that the front-end, not the contract, is your regulated surface. The Galeotti standard protects code, not operations.
• For interface operators: assume you are a regulated surface. Screen against sanctions lists, keep a token-listing policy, geo-block where you must, and document the compliance program, because being just a website is no longer the defense it once was.
• For users: know which layer can reach you. Your stablecoins can be frozen by their issuer regardless of the protocol; your front-end can screen your address; your on-chain history is permanent and analyzable; and self-custody protects your keys, not your anonymity.

The honest summary of DeFi compliance in 2026 is that the perimeter has moved. It is no longer a wall around an institution; it is a set of pressure points distributed through the stack, at the settlement asset, the interface, the whitelist, the credential, and the analytics feed. Whether that adds up to compliant DeFi or merely to DeFi with more choke points is the argument the industry will have for the rest of the decade.

Frequently Asked Questions

Is using DeFi legal in 2026?

Yes. Using decentralized finance protocols is legal in the United States and the European Union; there is no rule against interacting with a smart contract or self-custodying assets. What is regulated is the activity around a protocol: operating an interface that moves user funds, running an unlicensed money-transmitting business, transacting with sanctioned addresses, or failing to report taxable gains. Ordinary users owe tax on realized gains and should avoid sanctioned counterparties, but participation itself is not illegal.

Can a truly decentralized protocol be forced to do KYC?

Not directly. A protocol with no controller has no entity to serve an order on, which is the point of FATF’s control test: where no one exercises control or sufficient influence, the arrangement falls outside the VASP rules. Compliance instead attaches at the points that do have an operator: the front-end interface, the stablecoin issuer, permissioned pools, and credential providers. That is why 2026 compliance is built into the layers around a protocol rather than into the protocol itself.

What is the difference between the SEC and FinCEN in DeFi?

They police different things. The SEC regulates whether a token is a security and whether trading or brokering it requires registration. FinCEN, with OFAC, enforces anti-money-laundering and sanctions rules under the Bank Secrecy Act, and this is the surface most DeFi enforcement actually runs through. A common error is to describe money-laundering enforcement as an SEC crackdown; in the United States, AML authority sits with the Treasury, not the SEC.

Can stablecoins be frozen inside a DeFi position?

Yes. Major dollar stablecoins like USDT and USDC are issued by companies that can freeze, blacklist, or burn the balance at a specific address, and the GENIUS Act now requires payment-stablecoin issuers to hold the technical capability to comply with lawful freeze or seize orders. Because the freeze happens at the token level, it works regardless of how decentralized the protocol holding the stablecoin is; a frozen balance strands any DeFi position built on it.

What are Privacy Pools, and are they legal?

Privacy Pools is a system that uses zero-knowledge proofs to let users prove their funds do not come from known illicit sources without revealing their identity, an attempt to offer privacy without becoming a laundering tool. It is technically live on Ethereum, but its legal status is unsettled: it is not clear whether curating the required association sets amounts to censorship, or whether facilitating private transfers still counts as money transmission under the reading used to prosecute the Tornado Cash developer. Treat it as an active experiment, not a settled safe harbor.

Priya Reddy covers regulation and market structure for HOGE Wire.