Tornado Cash Developer Liability Across Jurisdictions
Prepared by the editorial team. Updated August 31, 2026.
Research Notice: This guide is part of our fintech research series examining blockchain privacy tools and their regulatory context. It is informational and educational only, is not legal, financial or compliance advice, and does not endorse or instruct the use of any mixing service. Laws differ by jurisdiction and change over time; verify current rules for your location.
Tornado Cash generated criminal proceedings in two countries, and they are routinely discussed as though they were one story with one verdict. They were not. A Dutch court and a New York federal court applied different statutes, faced different records and answered different questions, which is why the outcome in either place leaves the other exactly where it stood.
Why did two countries end up asking different questions?
Because criminal liability is defined by national statutes, and no two systems carve up conduct the same way. The same set of facts can satisfy the elements of an offense in one country and fail to match any offense in another, not because one system is more permissive but because the categories were drawn differently.
The clearest illustration in this matter is the count on which a New York jury returned a guilty verdict in August 2025: conspiracy to operate an unlicensed money transmitting business. That offense belongs to a US federal licensing scheme for money services, a regulatory architecture with no exact counterpart in Dutch law. A prosecutor in the Netherlands could not have charged it, and would have had to reach for entirely different provisions to address the same activity.
Procedure diverges as much as substance. The Netherlands tries serious cases before professional judges who issue a reasoned written judgment, while the US federal system uses a lay jury that returns a verdict without explaining itself. A reasoned judgment can be read, criticized and cited for its logic. A jury verdict tells you only which counts were proved to that jury’s satisfaction, which makes cross-border comparison harder than it looks.
What was decided in the Netherlands?
A Dutch court convicted Alexey Pertsev in May 2024 and imposed a sentence of 64 months. He was conditionally released in February 2025 to prepare an appeal, and that appeal remains pending. Until an appellate court rules, the first-instance judgment is the operative outcome but not the final word.
Two details are dropped from almost every English-language summary. The first is that conditional release pending appeal is a procedural step, not a reversal, and it says nothing about the merits. The second is that the appeal is live, so anyone describing the Dutch position as settled is describing an unfinished proceeding as a finished one.
The Dutch judgment is also a document rather than a headline. It sets out findings of fact, the provisions applied and the reasoning that connects them, and anyone seriously interested in what was decided should work from that text or a careful translation of it. Secondary coverage compresses a long reasoned judgment into a sentence, and the compression is where the errors enter.
Context also matters for how much the Dutch outcome can be generalized. The Netherlands applies national criminal law shaped by European anti-money-laundering instruments, so the analysis available to a Dutch court differs from what a court in a non-European system would have. Even within Europe, implementation varies between member states, and a first-instance judgment in one country does not announce a European rule.
What is the US case actually about?
It is a federal criminal prosecution in the Southern District of New York against Roman Storm. In August 2025 a jury convicted on one count, conspiracy to operate an unlicensed money transmitting business, and deadlocked on two others, money laundering conspiracy and sanctions evasion conspiracy. A deadlocked count is not an acquittal, so those counts remain open.
The case is unfinished in more than one respect. Storm moved for acquittal in September 2025, oral argument on that motion was heard in April 2026, and the motion remains undecided. A retrial on the two hung counts is scheduled for April 26, 2027 before Judge Katherine Polk Failla at the Thurgood Marshall Courthouse, with a final pretrial conference on April 20, 2027. Reporting indicates that conviction on both retrial counts could carry a maximum of up to 40 years.
It is worth separating this from the sanctions track entirely. Treasury removed Tornado Cash from the sanctions list in March 2025, and the criminal proceeding carried on regardless, because a listing decision and a criminal charge are different instruments with different tests. A delisting does not narrow an indictment, and a conviction does not restore a listing.
How can you compare how two jurisdictions frame the same conduct?
You write the conduct out as neutral facts, obtain the charging instrument in each country, match each fact to the elements of each offense, note the questions each court was never asked, and take the comparison to qualified counsel in both places. This is a research method, not legal advice.
Step 1: Write the conduct out in neutral facts
Describe what was actually done in plain language before attaching any legal label, because labels such as laundering or transmitting already carry the conclusion you are trying to test. Neutral facts also survive translation between systems, whereas labels rarely do.
Step 2: Find the charging instrument in each country
Obtain the indictment, summons or judgment that names the specific offenses in each jurisdiction, since coverage tends to describe a case by its subject rather than by the counts charged. The counts are the case, and everything else is context.
Step 3: Match each fact to the elements of each offense
List the elements a prosecutor must prove for each offense and mark which of your neutral facts would go to which element, because that is where two systems visibly diverge. Facts that carry an entire count in one country may be irrelevant in the other.
Step 4: Note what each court was never asked
Record the questions absent from each proceeding, since a court that was not asked about a point has not resolved it and cannot be cited as though it had. Silence in a judgment is not agreement with anything.
Step 5: Have qualified local counsel check each jurisdiction
Take the comparison to a qualified lawyer admitted in each country concerned, because statutory text read in translation misses procedure, precedent and prosecutorial practice. A comparison you assembled yourself is a starting point for a conversation, not a conclusion.
Does an outcome in one country settle the position in another?
No. A judgment binds within the legal system that issued it, and a court in another country is applying its own statutes to its own record. A conviction abroad is not proof of guilt at home, and an acquittal abroad is not a defense at home, though either may be relevant background in ways local counsel would assess.
This cuts both ways, and both directions are misreported. A guilty verdict in New York does not establish that a developer in Europe has broken European law. The Dutch judgment does not establish anything about US money transmission licensing, a subject Dutch law has no occasion to address. Each proceeding is evidence of how one system treated one defendant on one record.
The practical position for an open-source contributor is therefore uncertainty rather than a rule. What is reasonably clear is that the questions courts have engaged with concern control, revenue, promotion and knowledge of use, rather than authorship as such. What is not clear is where the line falls, and nobody should treat either case as marking it.
That uncertainty has practical consequences that go beyond the two defendants. Contributors weighing whether to work on privacy tooling cannot get a reliable answer from either proceeding, because neither produced a general standard and both are still moving. The responsible course is to treat the area as unsettled, to document decisions about design and revenue as they are made, and to take specific plans to a lawyer in the relevant country rather than reasoning from a headline about someone else’s case.
The two proceedings side by side
Setting the two matters against each other makes the divergence concrete, because the differences are structural rather than a matter of degree. The table summarizes publicly reported features of each proceeding. It is a research aid rather than legal advice, and it describes cases that are not finished.
| Feature | Netherlands | United States |
|---|---|---|
| Defendant | Alexey Pertsev | Roman Storm |
| Decision maker | Professional judges, reasoned judgment | Lay jury, verdict without reasons |
| Outcome to date | Convicted May 2024, 64 months | Guilty on one count, jury hung on two |
| Current posture | Conditionally released, appeal pending | Acquittal motion undecided, retrial set |
| Next scheduled step | Appellate proceedings, no public date | Retrial April 26, 2027 |
| Effect on the other country | None as a matter of law | None as a matter of law |
The last row is the one most often missed. Neither proceeding is a precedent for the other, and a reader who merges them into a single global verdict has invented an outcome that no court has delivered.
Frequently asked questions
Can the same person be prosecuted in two countries over the same conduct?
It is possible in principle, because separate sovereigns apply their own criminal law and protections against double jeopardy usually operate only within a single system. Treaties and prosecutorial discretion often limit this in practice, and anyone facing the question needs counsel in both places rather than a general rule.
Does living outside the United States remove US exposure?
Not by itself. US authorities can assert jurisdiction where there is a sufficient connection, such as US users, US infrastructure or dollar transactions, and extradition arrangements exist with many countries. Residence is one factor in an analysis rather than an answer to it.
Do these cases mean open-source contributors are generally at risk?
No general rule has emerged from them. Both proceedings turned on detailed findings about control, revenue, promotion and knowledge rather than on the act of contributing code, and neither produced an appellate statement about contributors at large. The position remains unsettled.
What happened to the other co-founder named in 2023?
Treasury designated Roman Semenov, a co-founder, in August 2023. A designation is an executive listing decision rather than a court verdict, so it carries no finding of guilt and should not be described as one.
