What is a money transmitter? The definition on trial

Two developers went to prison-facing trials over four words in a 1970s-era framework: what counts as transmitting money. The answer decides whether writing DeFi code is a regulated financial business or protected publishing, and Section 604 of the CLARITY Act is Congress’s attempt to settle it.

Most legal categories in crypto are abstractions until they are not. The money transmitter became concrete on the day federal agents arrested the developers of a piece of privacy software, charged them with operating an unlicensed money transmitting business, and put the crypto industry on notice that the government’s theory of a 1970s-era definition now reached people who wrote code and never touched a customer’s coin. The category has been the quiet workhorse of American crypto regulation for a decade, it is why exchanges hold 49 state licenses, why kiosks register with FinCEN, why every custodial app runs an anti-money-laundering program, and it has become, through the Tornado Cash and Samourai Wallet prosecutions, the sharpest legal question in the industry: can you transmit money you never control? Section 604 of the CLARITY Act, the provision law-enforcement lobbies are working hardest to narrow, is Congress’s attempt to answer by statute. This guide explains the category from its telegraph-era origins to its current trial: what a money transmitter is, who clearly is one, the control question that broke the consensus, and what the pending legislative answer would and would not change.

The category and its machinery

A money transmitter is, in the working federal definition, a person or business that provides money transmission services: accepting currency, funds, or other value that substitutes for currency from one person, and transmitting it to another location or person by any means. The definition’s breadth is deliberate. It was built to cover Western Union and its descendants, wire services, remittance shops, payment processors, any business whose product is moving other people’s money, and the phrase “other value that substitutes for currency” is the hinge that later swung crypto inside.

Being a transmitter carries two distinct regulatory loads, and confusing them confuses everything downstream. The first is federal: under the Bank Secrecy Act of 1970, as expanded by the Patriot Act, money transmitters are a category of money services business, MSB, regulated by FinCEN, the Treasury bureau that runs American anti-money-laundering law. An MSB must register with FinCEN, build and maintain an AML compliance program, file suspicious activity and currency transaction reports, keep records, and comply with sanctions administered by OFAC. The BSA’s design is deputization: financial institutions serve as the surveillance and reporting layer of law enforcement, and transmitters are drafted into that service. The second load is state: forty-nine states, all but Montana, separately require money transmitter licenses, each with its own application, bonding, capital, and examination regime, which is the notorious state-by-state maze that costs a national crypto business years and millions to assemble, and which federal charters and passage of preemptive legislation are perennially pitched as escaping.

Behind both sits the enforcement hammer that gives the category its teeth: 18 U.S.C. Section 1960, which makes operating an unlicensed money transmitting business a federal crime. Section 1960 is why the definition’s boundary is not an academic question. Classification as a transmitter without licenses is not a compliance gap to remediate; it is an indictment waiting for a prosecutor’s theory, and the statute’s reach, covering businesses that fail state licensing, fail FinCEN registration, or transmit funds known to be criminally derived, makes it the charge of choice in crypto cases.

Who is clearly inside

For most of the crypto stack, the classification analysis is settled and has been since FinCEN’s foundational 2013 guidance, which declared that administering or exchanging virtual currency is money transmission like any other: virtual currency and fiat are subject to the same rules.

Exchanges are transmitters: they accept value from customers and transmit it, between users, between currencies, out to wallets. Custodial wallet providers are transmitters, because holding users’ keys and moving funds at their instruction is the definition performed literally. Payment processors that accept crypto on behalf of merchants, over-the-counter desks, and crypto kiosk operators, the ATM networks whose compliance failures have made them a fixture of enforcement actions and, in the pending CLARITY text, the subject of first-ever federal operating standards, all inside. The practical consequence is the compliance architecture users experience without naming it: identity verification at onboarding, transaction monitoring, withdrawal reviews, the entire know-your-customer apparatus, which exists because the BSA requires it of financial institutions and the transmitter classification makes these businesses financial institutions.

It is worth pausing on how uncontroversial this half of the map is. The industry litigates many things; the proposition that an exchange holding customer funds is a regulated transmitter is not one of them. The war is entirely at the other edge of the category, where the software does the transmitting and no business ever holds the money.

The control question

The consensus, while it held, rested on a 2019 FinCEN guidance document that the industry treated as its constitution. Synthesizing years of interpretation, the guidance was widely read to say that what makes a transmitter is independent control over the value being moved: a business that accepts and holds funds, then sends them on, transmits money; a developer who publishes software through which users move their own funds, keys in their own hands, does not. Non-custodial wallets, decentralized exchange protocols, mixing software, on that reading, their creators were publishers of tools, not operators of financial businesses, and a decade of DeFi was built on the distinction.

The prosecutions broke it. The federal cases against the developers of Tornado Cash, the Ethereum privacy protocol, and Samourai Wallet, the Bitcoin privacy wallet, advanced under Section 1960 a theory that horrified the industry precisely because of what it did not require: custody. The government’s position, as the industry’s lawyers characterized it, defined a new class of money transmitting entities, developers of decentralized protocols where no intermediary ever controls user tokens at any stage. If writing and maintaining software that moves value is transmission, regardless of who holds the keys, then the 2019 line does not exist, and every non-custodial developer in the country carries latent criminal exposure that turns on prosecutorial discretion. Legal scholars, including counsel writing in the Stanford Blockchain Review, framed the resulting question in its cleanest form: can you transmit money you never control? The industry’s answer is no, and that the contrary theory converts code publication, an activity with First Amendment dimensions, into an unlicensed financial business. Law enforcement’s answer is that the question is too clever by half: the developers built, deployed, updated, and in some tellings profited from machines whose function was moving money, much of it criminal, and the custody test is a formalism that launders responsibility. Courts have not settled it. The cases produced plea agreements, contested rulings, and doctrine that remains genuinely open, which is the worst possible state for an industry deciding where to build: the boundary of a felony is currently a litigation position.

The state layer, and the maze beneath the maze

Before the statutory answer, one more layer of the map deserves its own treatment, because the federal definition this guide has traced is only half of any crypto business’s transmitter problem, and often the cheaper half.

Money transmission is regulated on two independent tracks, and the state track came first. Forty-nine states plus territories license transmitters under their own statutes, each with its own definition of transmission, its own exemptions, its own capital, bonding, and net-worth requirements, and its own examiners. The definitions do not match: an activity that is transmission in one state is exempt in another, some states carve out closed-loop systems or agent-of-payee arrangements, and several have enacted bespoke virtual-currency regimes on top, New York’s BitLicense the most famous, with its own application economics measured in years and legal fees in the millions. A national crypto business therefore does not ask whether it is a money transmitter; it asks the question up to fifty times, against fifty texts, and maintains the resulting license portfolio through fifty renewal and examination cycles. Industry estimates of the full build have run from the high seven figures into eight, before a single federal obligation attaches.

The maze’s existence explains three otherwise puzzling features of the industry’s structure. It explains the partnership model, fintechs and crypto apps riding licensed sponsors’ transmitter licenses instead of acquiring their own, an arrangement whose fragility the Synapse collapse exposed from the banking side. It explains the outsized value of any federal instrument that preempts the states: the OCC trust charters at the center of the current bank-lobby fight are prized precisely because a federal charter can replace the fifty-license portfolio, which is also why state regulators, through the Conference of State Bank Supervisors, oppose them alongside the banks. And it explains a persistent asymmetry in Section 604’s politics: the provision addresses only the federal BSA definition, and a developer shielded federally could in principle still face state transmitter theories, though the practical center of gravity, and the criminal exposure that made the prosecutions existential, is federal, which is why the statutory fight concentrates there.

The state layer also supplies the definitional cautionary tale the federal debate should study. Uniformity projects, model acts, multistate licensing agreements, examination passporting, have worked at the state level for years and closed perhaps half the gap, while the underlying definitional divergence persists, because each legislature guards its own text. The lesson for readers of Section 604 is direct: definitions of money transmission have never converged voluntarily, at any level of American government, in fifty years of trying. They converge when a superior authority writes one text that binds everyone, which is what the CLARITY provision would be, and why a paragraph of statutory language is worth more, to both sides of the fight, than a decade of guidance ever was.

Section 604, the statutory answer

This is the dispute Section 604 of the CLARITY Act exists to end, and understanding the provision means understanding both what it does and the fight over its edges.

The section, descended from the Blockchain Regulatory Certainty Act that was folded into the House bill, draws the line in statute where the 2019 guidance drew it in interpretation: developers and publishers of non-custodial software, code that never takes control of user funds, are not money transmitters under the Bank Secrecy Act by virtue of publishing or maintaining it. The custody test becomes law rather than guidance, retroactively vindicating the industry’s decade-old reading and prospectively closing the Section 1960 theory the privacy prosecutions ran on. For DeFi, the provision is close to existential, which is why the industry’s lobbying has treated its preservation in the merged Senate text as a red line.

The opposition is institutional and specific: the National District Attorneys Association, joined by sheriffs’ and prosecutors’ organizations, argues the provision would materially impair criminal investigations by severing the liability connection between protocol developers and the financial activity their code enables, removing the charge that lets investigators reach the infrastructure layer of laundering operations. Senator Wyden’s response compresses the other side to a sentence: developers who never control customer funds should not be classified as money transmitters for publishing code. The negotiation between those positions, conducted through Senator Cortez Masto’s drafting sessions, is where the provision’s real content will be set, and the things to watch are the carve-backs: language distinguishing mixers from wallets, front-end operators from contract deployers, or profiting maintainers from mere publishers would mark the compromises law enforcement extracted.

Equally important is what Section 604 does not touch. Custodial businesses remain fully inside the perimeter: exchanges, hosted wallets, processors, and kiosks keep every BSA obligation and every state license, and the CLARITY framework separately extends bank-secrecy duties across registered digital-asset intermediaries. The provision does not deregulate crypto’s financial businesses; it declines to regulate its publishers, which is a distinction the debate on both sides has incentives to blur. For anyone building or holding in this industry, the practical summary is three sentences long. If a business holds user funds, it is a money transmitter, licensed and surveilled, and nothing pending changes that. If software never holds them, its authors’ status is today a contested prosecution theory and would tomorrow, under 604, be statutory safety. The distance between those two sentences is where two developers stood trial, and where American DeFi’s legal future is being drafted this month.

One historical footnote completes the picture and explains the category’s peculiar gravity in crypto enforcement: Section 1960 was itself a product of a moral panic about new money technology. Congress enacted it in 1992, aimed at storefront wire services suspected of laundering drug proceeds, and strengthened it in the Patriot Act era, when the removal of a knowledge requirement for state-licensing violations converted it into something close to a strict-liability trap: a business that misjudges whether a state considers it a transmitter commits a federal crime by operating, whatever it believed. That structure, born decades before anyone imagined non-custodial software, is what gives the current definitional fight its stakes. Most regulatory misclassifications in finance produce deficiency letters and fines; misclassification under 1960 produces indictments, and the statute has become the government’s most flexible crypto charge precisely because its elements are so spare, no fraud required, no victims required, only transmission without license. The industry’s decade of pleading for regulatory clarity has always been, at bottom, a plea about this statute: not for permission to operate, which custodial businesses obtain through licensing, but for certainty about which side of a criminal line an activity sits on. Section 604 is the first legislative text that would draw that line in statute rather than guidance, which is why a provision that changes no license requirement for any operating business has nonetheless become one of the most fought-over paragraphs in the bill.

Disclaimer: This article is for information and educational purposes only and does not constitute financial, investment, or legal advice. It describes statutes, guidance, active litigation, and pending legislation whose interpretation and status can change. Anyone facing classification questions should consult qualified counsel. Always do your own research. Information is accurate as of July 21, 2026.

Frequently Asked Questions