Uniswap v4 Hooks: Legal Implications for Protocol Developers (2026 Guide)

Analysis for CryptoLaw.blog — DeFi & Protocols

When Uniswap v4 went live on Ethereum mainnet on January 31, 2025, the legal conversation around decentralized exchanges quietly changed categories. v2 and v3 were products: a fixed constant-function market maker with a small set of fee tiers. v4 is a platform. Its singleton PoolManager exposes lifecycle callbacks that any developer can implement in an arbitrary contract — a “hook” — and attach to a pool at initialization. That architectural shift matters legally because it relocates discretion. In v3, no one decided what happened during your swap. In v4, someone wrote code that decides what happens during your swap, and that someone may have a name, a wallet, a revenue stream, and a U.S. address.

This piece is written for the people actually shipping hooks and the counsel advising them. The thesis is straightforward: the hook, not the AMM, is now the locus of regulatory risk, and the analysis turns almost entirely on three variables — whether the hook touches user funds, whether it exercises ongoing discretion, and whether it pays someone.

What a Hook Actually Is (And Why the Architecture Is a Legal Fact)

v4 replaces v3’s factory-and-many-pools model with one contract holding all pool state, plus flash accounting via EIP-1153 transient storage. Each pool is identified by a PoolKey struct that includes the two currencies, the fee, the tick spacing, and — critically — the address of the hook contract.

A hook advertises which callbacks it implements through the low bits of its own address. The Hooks library in v4-core defines a 14-bit permission mask:

uint160 internal constant BEFORE_INITIALIZE_FLAG            = 1 << 13;
uint160 internal constant BEFORE_ADD_LIQUIDITY_FLAG         = 1 << 11;
uint160 internal constant BEFORE_SWAP_FLAG                  = 1 << 7;
uint160 internal constant AFTER_SWAP_FLAG                   = 1 << 6;
uint160 internal constant BEFORE_SWAP_RETURNS_DELTA_FLAG    = 1 << 3;
uint160 internal constant AFTER_SWAP_RETURNS_DELTA_FLAG     = 1 << 2;
// ...14 flags total

Because permissions are encoded in the address, developers must mine a CREATE2 salt (the HookMiner utility in v4-periphery exists for exactly this) to deploy at an address whose trailing bits match the intended permission set. The legal consequence is underappreciated: a hook’s permission surface is cryptographically fixed at deployment and cannot be expanded without deploying to a new address. That is a genuinely useful evidentiary fact. A developer who deploys a hook whose address lacks the RETURNS_DELTA flags can demonstrate, from chain state alone, that the contract is structurally incapable of altering swap economics or extracting value from a trade. In a regulatory inquiry, that is a much stronger showing than a whitepaper assurance.

The caveat: address bits constrain which callbacks fire, not what the callback body does. A hook deployed behind a proxy with a delegatecall to an upgradeable implementation has fixed permissions and entirely mutable logic. Regulators and plaintiffs will read the admin key, not the address.

The most legally consequential flags are the delta-returning ones. A hook with BEFORE_SWAP_RETURNS_DELTA can substitute its own pricing for the pool’s curve:

function beforeSwap(
    address sender,
    PoolKey calldata key,
    SwapParams calldata params,
    bytes calldata hookData
) external returns (bytes4, BeforeSwapDelta, uint24);

When that BeforeSwapDelta is non-zero, the hook is no longer observing a trade. It is filling one. That distinction — passive observer versus principal counterparty — is the fault line running through every analysis below.

Does Writing a Hook Make You an Intermediary?

Money transmission and 18 U.S.C. § 1960

FinCEN’s foundational 2019 guidance (FIN-2019-G001) draws the line at independent control over value: a person who merely develops or distributes software is not a money transmitter, but a person who accepts and transmits value on another’s behalf is. Applied to v4, most hooks pass. A dynamic-fee hook that reads an oracle and returns a fee override never custodies anything; the PoolManager holds the tokens.

The exceptions are the hooks that hold. A hook implementing on-chain limit orders, a TWAMM, or a yield strategy typically escrows user tokens in the hook contract itself between blocks. If that contract has an owner key capable of moving those balances, the “independent control” test starts to bite, and the developer is arguably operating a money transmitting business — triggering FinCEN MSB registration under 31 C.F.R. § 1022.380 and state licensing.

The stakes here are not hypothetical. In United States v. Storm (S.D.N.Y.), the jury returned a verdict in August 2025 convicting the Tornado Cash developer on a conspiracy count under 18 U.S.C. § 1960 while hanging on the money laundering and sanctions counts. Whatever one thinks of the outcome, the message to builders was unambiguous: writing and deploying immutable financial software is not, by itself, a complete defense. The mitigating counterweight is DOJ’s April 7, 2025 memorandum from Deputy Attorney General Todd Blanche (“Ending Regulatory Weaponization Against Digital Assets”), which disbanded the National Cryptocurrency Enforcement Team and directed prosecutors not to charge § 1960 violations absent evidence the defendant knew the business was unlicensed. That is a charging policy, not a statute — it binds this Department, not the next one, and not a state attorney general.

Exchange, broker-dealer, and dealer status

The SEC’s 2022/2023 proposal to expand Exchange Act Rule 3b-16 to cover “communication protocol systems” — the rule most directly aimed at DeFi front-ends and protocols — was formally withdrawn in June 2025. Separately, the SEC’s February 2024 dealer rule was vacated in Crypto Freedom Alliance of Texas v. SEC (N.D. Tex., Nov. 21, 2024) on the ground that the Commission exceeded its statutory authority. Both withdrawals meaningfully reduced the near-term risk for hook developers running liquidity strategies.

“Reduced” is not “eliminated.” Statutory dealer analysis under Exchange Act § 3(a)(5) survives the rule’s vacatur. A delta-returning hook that systematically quotes both sides of a market in a token that is itself a security, for the hook operator’s own account and as part of a regular business, looks like a dealer under pre-existing law. The token classification is the gate: if the pool trades ETH/USDC, the securities analysis largely evaporates. If it trades a token with a live Howey problem, the hook operator inherits that problem.

CFTC jurisdiction

The CFTC has been the more aggressive DeFi regulator on protocol design, and its September 7, 2023 settled actions against Opyn, ZeroEx (0x), and Deridex are the closest analogue to hook risk in the existing enforcement record. The Commission charged those developers with offering illegal off-exchange leveraged retail commodity transactions under CEA § 2(c)(2)(D), failing to register as a DCM/SEF or FCM, and failing to implement a customer identification program. Notably, ZeroEx’s liability arose from a protocol enabling access to leveraged tokens — a permissioning and routing function conceptually similar to what a hook does. Commissioner Summer Mersinger’s dissent, arguing the Commission was regulating by enforcement against software developers, remains the best articulation of the builder-side counterargument.

Practical translation: a hook that enables margin, perpetual-style funding rates, or any position with leverage delivered to retail U.S. persons is squarely in CFTC territory regardless of how decentralized the underlying AMM is. CFTC v. Ooki DAO (N.D. Cal., default judgment June 2023) confirmed the agency will pursue an unincorporated collective as an unincorporated association when no entity presents itself.

Howey and the Hook Token

Most hook developers do not intend to issue a security. Many do it accidentally, through the fee mechanism.

Consider the common pattern: a hook takes a spread on each swap via afterSwapReturnDelta, routes a portion to a treasury, and distributes treasury proceeds pro rata to holders of a “hook token” that also governs parameter changes. Run SEC v. W.J. Howey Co., 328 U.S. 293 (1946):

• Investment of money — satisfied by purchase in the secondary market; Uselton v. Commercial Lovelace Motor Freight confirms consideration need not be cash.
• Common enterprise — horizontal commonality is straightforward where all holders share one fee pool.
• Expectation of profits — the fee distribution is the profit. Marketing language about “real yield” is the exhibit the staff will attach.
• Efforts of others — the decisive prong. Under SEC v. Glenn W. Turner Enterprises, 474 F.2d 476 (9th Cir. 1973), the question is whether the efforts are “undeniably significant” and essential to the enterprise’s success. An immutable, parameterless hook with a fee stream is a genuinely hard case for the SEC. An upgradeable hook whose team tunes curves, adds pools, and manages a treasury is not.

Two 2025 SEC developments cut in builders’ favor. The Division of Corporation Finance issued a staff statement on protocol staking activities (May 29, 2025) and a follow-on addressing liquid staking (August 2025), both reasoning that administrative or ministerial activity does not supply the entrepreneurial effort Howey requires. The framing is portable to hooks: ministerial execution of predetermined logic is not managerial effort. Builders who want that argument should be able to point to code that makes it true — no admin key, no discretionary parameters, no treasury.

Also relevant context: the SEC closed its Uniswap Labs investigation without enforcement in February 2025, after issuing a Wells notice in April 2024, and Chairman Paul Atkins announced “Project Crypto” in July 2025 signaling a move toward rulemaking over enforcement. This is a policy posture, not a safe harbor.

Protocol vs. Front-End: Still the Most Important Distinction

The single most useful precedent for hook developers is Risley v. Universal Navigation Inc., 690 F. Supp. 3d 236 (S.D.N.Y. Aug. 29, 2023), where Judge Katherine Polk Failla dismissed claims against Uniswap Labs arising from scam tokens traded on the protocol, analogizing the developers to a manufacturer of a product misused by third parties. The court declined to hold protocol authors liable for the conduct of token issuers transacting through neutral code.

That reasoning maps onto v4 with a wrinkle. In v4, the hook is the customization that Risley treated as third-party conduct. The PoolManager and the Uniswap Labs interface occupy the neutral-infrastructure position; the hook developer occupies the position of the party whose specific design choices caused the harm. Hook developers should not assume they inherit Risley’s protection — in the v4 stack, they are more often the defendant it protects others from.

On the front-end side, the Treasury/IRS “DeFi broker” regulations finalized December 27, 2024 — which would have imposed Form 1099-DA reporting on trading front-end service providers — were nullified under the Congressional Review Act, signed April 10, 2025. That repeal removed the most concrete federal reporting obligation aimed at interfaces. It did not repeal anything applicable to the protocol layer, because nothing there existed to begin with.

The operative rule for builders: keep the interface and the contracts in separate legal and technical containers. If your hosted UI charges a routing fee, curates a pool allowlist, or performs geofencing, that is an intermediary-flavored business and should sit in an entity that can hold licenses. The hook contracts should be deployable and usable without it.

Sanctions, Compliance Hooks, and the Van Loon Problem

v4 made permissioned pools trivial. A beforeSwap hook can check an attestation registry and revert for unverified addresses, producing a KYC-gated pool on a permissionless AMM. Institutional adoption of tokenized assets has driven real demand for this pattern.

The sanctions backdrop shifted materially. In Van Loon v. Department of the Treasury, 122 F.4th 549 (5th Cir. Nov. 26, 2024), the Fifth Circuit held that Tornado Cash’s immutable smart contracts are not “property” within the meaning of IEEPA and therefore could not be blocked; OFAC delisted the addresses on March 21, 2025. Read narrowly, Van Loon stands for a proposition hook developers should internalize: immutability is legally load-bearing. A contract no one can control is harder to characterize as blockable property or as an ownable service.

But immutability and compliance are in genuine tension. An address-screening hook with an updatable blocklist is, by construction, controlled — and control is what OFAC, FinCEN, and the SEC all look for. There is no clever way around this. Builders should choose deliberately: either build an immutable, ungated hook and accept that you cannot screen, or build a gated hook and accept intermediary-style obligations, including OFAC screening against SDN and, where applicable, NS-MBS listings.

Entity Structure and Liability for the Hook Team

Unwrapped hook collectives face general partnership exposure. Sarcuni v. bZx DAO (S.D. Cal. 2023) allowed claims to proceed on the theory that DAO participants were general partners jointly and severally liable, and Samuels v. Lido DAO (N.D. Cal., Nov. 18, 2024) held that Lido DAO could plausibly be a general partnership with institutional token holders as partners. For a hook earning fees and governed by token vote, that is the default characterization if nothing else is filed.

Realistic options: a Delaware or Wyoming LLC for the development entity (simplest, appropriate for most teams); Wyoming’s Decentralized Unincorporated Nonprofit Association statute, W.S. §§ 17-32-101 et seq., effective July 1, 2024, for genuinely member-governed protocols seeking liability shielding without a member registry; or a Cayman or Marshall Islands foundation where a non-U.S. protocol entity is defensible on facts. Note that Wyoming’s earlier DAO LLC statute drew criticism for its public-membership and algorithmic-management provisions; the DUNA is generally the better-drafted vehicle.

On bug liability: audits are risk mitigation, not immunity. A hook that mishandles the BalanceDelta accounting can drain the shared PoolManager flash-accounting context. Standard exculpatory language in a UI’s terms of service does not reach a user who never visited that UI. Protocol coverage from Nexus Mutual or a comparable provider, plus tech E&O for the development entity, is the practical answer — and note that most crypto-native coverage requires a completed audit as a condition precedent.

State and International Overlay

New York. NYDFS treats virtual currency business activity broadly under 23 NYCRR Part 200. A hook that custodies user assets and is promoted to New York residents raises BitLicense exposure regardless of federal posture.

California. The Digital Financial Assets Law (Fin. Code §§ 3101 et seq.) reaches digital financial asset business activity with California residents, with a licensing regime administered by the DFPI. Its operative date was pushed back by AB 1934 to July 1, 2026 — verify current status and any DFPI transition relief before relying on this. Custodial hooks marketed to California users should be assessed under it.

Federal legislation. The House passed the Digital Asset Market Clarity (CLARITY) Act on July 17, 2025 by a 294–134 vote. Its provisions would allocate spot digital commodity authority to the CFTC and — most relevant here — codify a non-custodial developer safe harbor drawn from the Blockchain Regulatory Certainty Act, protecting developers who do not control user funds. Senate action and enactment status as of this writing should be confirmed independently; nothing in this article assumes the bill is law.

EU/MiCA. Regulation (EU) 2023/1114 conditions its DeFi treatment on the absence of an intermediary (see Recital 22). A hook developer who takes a fee, controls an upgrade key, and operates a front-end serving EU users is not obviously outside CASP authorization, and ESMA has signaled that nominal decentralization will not carry the argument. Per CryptoLaw.blog’s editorial standard, the U.S. relevance is direct: U.S. hook teams routinely serve EU users through a single global interface, and the same admin key that creates SEC “efforts of others” exposure creates MiCA intermediary exposure.

Risk Assessment

Hook archetype	Primary exposure	Risk
Non-custodial analytics / event logging (afterSwap, no delta)	Effectively none	Low
Dynamic fee via oracle, no fee capture	Oracle manipulation tort claims	Low
Custom curve / delta-returning, immutable, no fee	Statutory dealer if securities traded	Low–Moderate
Fee-capturing hook funding a governance token	Howey; unregistered offering	High
Custodial limit-order / TWAMM with admin key	FinCEN MSB, § 1960, state MTL, DFAL/BitLicense	High
Leverage or perp-style funding to U.S. retail	CEA § 2(c)(2)(D); Opyn/ZeroEx/Deridex pattern	Very High
KYC/allowlist gating hook with updatable list	OFAC screening duties; control undercuts decentralization	Moderate–High
Hosted front-end with routing fee	Broker/intermediary characterization; state licensing	Moderate–High

The pattern is legible: risk scales with custody, discretion, and revenue — roughly in that order. Immutability without fees is close to a safe harbor in practice; upgradeability plus a revenue-sharing token is close to the opposite.

Key Takeaways for Builders

1. Mine the narrowest address you can. Every permission flag you omit is a permanent, chain-verifiable representation about what your contract cannot do. Do not request delta flags “for future flexibility.”
2. Custody is the bright line. If tokens rest in your hook between transactions and an admin key can move them, assume money transmitter analysis applies and budget for licensing counsel before launch — not after.
3. Choose immutability or compliance, explicitly. Van Loon rewards contracts nobody controls. Screening hooks require a controller. You cannot claim both postures.
4. Do not monetize through a token by default. The fastest route from “infrastructure” to “unregistered security” is a fee stream pointed at a tradable governance asset. Take fees to a corporate entity and pay taxes on them instead.
5. Separate the interface from the contracts. Different entity, different repository, different terms of service. The interface is where geofencing, screening, and disclosure live.
6. Wrap the team. Sarcuni and Samuels make unwrapped collectives a general partnership by default. A Delaware LLC costs a few hundred dollars; joint and several liability does not have a price.
7. Never offer leverage to U.S. retail. This is the one line with a settled, adverse enforcement record and no meaningful policy relief.
8. Document the deployment. Keep audit reports, the salt-mining record, and a contemporaneous memo on parameter immutability. Regulatory posture is proven with artifacts.
9. Check the license. v4-core ships under the Business Source License 1.1 with an Additional Use Grant. Confirm your use qualifies before building a commercial product on forked core code.

The regulatory environment in 2025–2026 has been more favorable to protocol developers than at any prior point — withdrawn rules, dropped investigations, a repealed broker regulation, and a Justice Department charging policy that narrows § 1960. None of that is codified. The Storm verdict remains a reminder that a jury, not a policy memorandum, decides what a developer is responsible for. Build as though the next administration reads your admin key first.

---

This content is for informational purposes only and does not constitute legal advice. Consult with a qualified US attorney for guidance specific to your situation.