The Bitcoin Fog Precedent: The DOJ's Quiet Attempt to Make Jurisdiction Follow the Code

The Bitcoin Fog Precedent: The DOJ's Quiet Attempt to Make Jurisdiction Follow the Code

Hook

On the last Friday of the filing window, the U.S. Department of Justice submitted a short supplemental authority notice in the Southern District of New York. It did not announce a new charge. It did not produce a witness. It simply told the court that the D.C. Circuit's ruling in United States v. Roman Sterlingov — the Bitcoin Fog case — "directly supports" its position that Manhattan has jurisdiction over Roman Storm.

That is the entire event. A citation. A procedural nudge filed under a docket number most retail traders will never open.

And yet this is the most consequential sentence written in crypto law this quarter. The DOJ is not arguing that Storm did something new in New York. It is arguing that a precedent built on a custodial, brick-and-mortar money transmission business — an operator whose customers physically visited a Washington D.C. office — can be transplanted wholesale onto a non-custodial smart contract that has no office, no server, and no legal entity anywhere on Earth. The code whispered truth; the balance sheet lied. Here, the ledger is a court filing, and the lie is that the two architectures are the same thing.

I have spent eleven years watching regulators try to fit decentralized systems into centralized statutes. This filing is the cleanest example yet of the strategy: pick the precedent that reaches the outcome, then argue the architecture is irrelevant.

Context: Two Mixers, One Docket

To understand why a citation matters, you have to understand what Bitcoin Fog was — and what Tornado Cash is not.

Bitcoin Fog launched in 2011 as one of the earliest Bitcoin mixing services. It was, by every technical measure, a custodial business. Users sent coins to an address controlled by the operator. The operator pooled, shuffled, and returned funds. There was a human being in the middle of every transaction, and that human being had an address. When an undercover agent completed transactions at the operator's office in Washington D.C., the government did not need to theorize about where the business operated. It walked in.

Roman Sterlingov was arrested in 2021, convicted in 2024 on money laundering and unlicensed money transmitting charges, and sentenced to prison. His appeal failed. The D.C. Circuit affirmed, holding that the District of Columbia had jurisdiction over all four counts because Bitcoin Fog had served customers in that district. That is the precedent the DOJ now waves at Judge Katherine Polk Failla.

Tornado Cash is a different animal entirely. It is a set of immutable smart contracts deployed on Ethereum. There is no operator account that holds user funds. There is no office. There is no server that can be seized. The contracts route deposits and withdrawals through zero-knowledge proofs that break the on-chain link between sender and receiver. The original team wrote code, deployed it, and walked away. The protocol keeps running because Ethereum keeps running. The smart contract does not care about your hopes, your jurisdiction, or your subpoena.

Roman Storm is a co-founder of that project. In August 2025, a federal jury in Manhattan convicted him on one count — unlicensed money transmitting — and deadlocked on the other two: money laundering conspiracy and sanctions evasion conspiracy. He then filed a post-conviction motion for a judgment of acquittal, arguing the evidence cannot support even the count he lost. The DOJ's response is where we are now.

There is a third jurisdiction in play. Alexey Pertsev, another Tornado Cash developer, was convicted in the Netherlands in 2024. That means the same protocol has now produced criminal convictions on two continents, from two legal systems, against two developers who wrote the same open-source code. That is not a coincidence. That is a coordination pattern, and it should terrify anyone who has ever pushed a commit to a permissionless protocol.

The DOJ's supplemental filing does not exist in a vacuum. It is the latest move in a multi-year campaign to establish that writing code can be a crime. The question is whether the courts will accept an analogy that collapses two fundamentally different technologies into one legal category.

Core: The Architecture Gap the Prosecution Wants You to Ignore

The DOJ's jurisdictional argument rests on what lawyers call minimum contacts — the constitutional principle that a court may exercise power over a defendant only if that defendant has purposefully established meaningful connections with the forum. In the Bitcoin Fog case, that connection was easy to establish. The business had customers in D.C. The operator had an office in D.C. The undercover agent made purchases in D.C. Minimum contacts were not a legal theory; they were a fact pattern with a street address.

The DOJ now wants to run the same play against Tornado Cash. Its evidence of Manhattan contact is thinner but clever: testimony from Shakeeb Ahmed, a convicted attacker, that he used Tornado Cash from his Manhattan apartment. Ahmed is not a sympathetic witness. He is a hacker who stole tens of millions from a DeFi protocol and then laundered the proceeds. But his testimony serves a narrow purpose: it puts the protocol's usage physically inside the Southern District of New York.

Here is the logical sleight of hand. If a protocol is used by anyone in a district, the DOJ argues, that district has jurisdiction over the people who wrote it. Under that theory, the location of the developer is irrelevant. The location of the code is irrelevant. What matters is where a single user happened to be sitting when they clicked a button.

This is not a jurisdictional doctrine. It is a jurisdictional dragnet. Extend it and every permissionless protocol is subject to the laws of every jurisdiction on the planet, simultaneously, because someone somewhere used it. A developer in Mexico City — and I write this as one — would be criminally exposed to Manhattan, Amsterdam, and any other district that decides to assert itself, based purely on the geography of strangers.

I traced the ghost liquidity back to its source in my own work, and I can tell you the source here is not the code. It is the analogy. The DOJ needs Tornado Cash to look like Bitcoin Fog because the Bitcoin Fog precedent only works if the two are equivalent. So let me do the forensic work the prosecution hopes the court skips.

Bitcoin Fog was a business. Tornado Cash is a protocol. A business has revenue, employees, contracts, and a physical footprint. A protocol has none of those things. When you arrest the operator of a business, you are arresting the entity. When you arrest the author of a protocol, you are arresting a person who is no longer in the loop. The operator of Bitcoin Fog could, at any moment, choose to stop the service, refuse a customer, or hand over records. The author of Tornado Cash cannot. The contracts execute whether the author lives or dies.

Bitcoin Fog held custody. Tornado Cash never did. This is the crux, and it is the part the DOJ's filing glosses over. Custody is the legal hinge of money transmission law. You are a money transmitter because you take control of someone's funds and move them. Bitcoin Fog took control. Tornado Cash cannot take control — the contracts have no owner key over user deposits. The user's funds move from their wallet to a pool and back out, governed by math, not by a human decision. The prosecution's entire money transmitting theory requires custody that the architecture does not provide.

Bitcoin Fog had a location. Tornado Cash has a hash. You can seize a server. You can raid an office. You cannot seize a deployed contract without convincing a validator set to rewrite a chain, which is precisely the kind of censorship the whole system was designed to resist. The DOJ's minimum contacts theory works for physical businesses because physical businesses have physical coordinates. Software does not.

This is not a technicality. It is the entire ballgame. If the court accepts the DOJ's framing, it will have effectively ruled that the architecture of a protocol is legally invisible — that a custodial mixer and a non-custodial one are the same thing because both move money. That ruling would not be about Tornado Cash. It would be a blanket statement that the law does not recognize the distinction between holding someone's keys and publishing code.

The Count That Actually Matters

Beneath the jurisdiction fight sits a more concrete and more dangerous precedent: the unlicensed money transmitting conviction. This is the one count the jury actually agreed on. It is the one count Storm is trying to overturn with his acquittal motion. And it is the one count that, if it survives appeal, will do the most damage.

Money transmission law was written for a world of money transmitters — Western Union, MoneyGram, the corner check-cashing shop. The statute assumes a business that receives and transmits funds on behalf of customers, subject to registration and reporting. Regulators have spent a decade trying to decide whether software developers fall inside that definition. The Storm conviction, if affirmed, answers yes.

Think about what that means in operational terms. A money transmitter must register, must implement AML programs, must file reports, must identify customers. If a developer can be convicted of unlicensed money transmission for deploying code, then every permissionless protocol author is potentially an unlicensed money transmitter — because their software moves value and they did not register. The registration requirement becomes impossible to satisfy, because there is no entity to register. The law demands a license that cannot be issued to a ghost.

This is the real precedent war. The jurisdiction question is procedural. The money transmitting question is existential. A jurisdiction ruling tells you where a developer can be tried. A money transmitting ruling tells you whether writing the code was the crime.

The DOJ understands this ordering. That is why it filed the Bitcoin Fog citation now, before the acquittal motion is heard. The strategy is layered: reinforce jurisdiction first, then defend the substantive count using the jurisdictional win as a foundation. If the court accepts that Manhattan has power over Storm, the money transmitting conviction becomes easier to sustain, and the deadlocked counts become easier to retry.

The Jury Split as a Signal, Not a Setback

I want to correct a misreading that has spread through crypto media. When the August 2025 jury deadlocked on money laundering and sanctions conspiracy, much of the commentary treated it as a partial victory for Storm — proof that the case was weak, that the government overreached, that the developer-defense narrative was winning.

Read the deadlock more carefully. A hung jury is not an acquittal. It is a signal that the evidence was close enough that reasonable jurors could not agree. In legal practice, a deadlock on some counts plus a conviction on another is not a loss for the prosecution. It is an invitation to retry. The government retains the option to bring the deadlocked counts again, and the DOJ's filing pattern suggests it is laying the groundwork to do exactly that.

The timeline reinforces this. Judge Failla scheduled the hearing on Storm's motion for April 2026. That is not a near-term event. That is a deliberate pause that keeps the entire matter — and the entire privacy sector — under a cloud of unresolved legal risk for more than a year. During that window, the government can continue to build jurisdiction, cite new authority, and wait for other courts to produce favorable rulings it can borrow.

Silence in the logs is louder than the hack. Here, the silence is the delay. Every month the case stays open, the regulatory overhang compresses the entire permissionless privacy segment. Developers do not know if their code is legal. Investors do not know if their holdings are exposed. Protocols do not know if they will be named next. The absence of a final ruling is itself a form of enforcement.

The Minimum Contacts Trap

The jurisdictional theory the DOJ is advancing deserves to be stated in its most aggressive form, because that is the form that will survive if unchallenged.

Under the government's logic, jurisdiction attaches wherever a protocol has users. Bitcoin Fog had users in D.C., so D.C. had jurisdiction. Tornado Cash had a user in Manhattan, so Manhattan has jurisdiction. The connecting principle is not where the defendant acted. It is where the technology was touched.

Apply that principle generally and the consequences are absurd. A decentralized exchange deployed by an anonymous team gets used by a trader in Tokyo, a farmer in Lagos, and a student in Buenos Aires. By the government's logic, all three jurisdictions — and every other place with a single user — can claim criminal jurisdiction over the developers. There is no limiting principle. The developer becomes subject to the sum total of all jurisdictions, defined by the geography of strangers they have never met.

American constitutional law has a doctrine that is supposed to prevent exactly this. Due process requires that jurisdiction be reasonable, not merely possible. A court must weigh the defendant's burden, the forum's interest, and the interstate — or international — system's interest in efficiency. The DOJ's approach flips the analysis: it treats any contact, no matter how incidental, as sufficient. That is not how minimum contacts is supposed to work. It is how a dragnet works.

I have audited contracts for teams scattered across four continents. Not one of them could tell you where their users are, because the entire point of a permissionless protocol is that you do not know and do not need to. That ignorance — which the government treats as evidence of wrongdoing — is the defining feature of the technology. If the courts accept that not knowing your users is a crime, they will have criminalized the architecture itself.

The D.C. Circuit's Bitcoin Fog ruling did not have to answer this question, because Bitcoin Fog's operator did know his customers. He served them in person. The precedent is being stretched past its facts. Whether the Southern District allows that stretch is the question of the year.

The Chilling Effect Nobody Is Pricing

The most underreported consequence of this case is not legal. It is behavioral.

When the core developer of a widely used protocol faces criminal conviction, the message to every other developer is immediate and unambiguous: the code you write today can send you to prison tomorrow. This is the definition of a chilling effect — the suppression of lawful activity through the fear of punishment. And in the privacy sector, it is already working.

Consider the incentive structure. A developer choosing where to build now faces a simple calculation. Write a permissionless privacy protocol and accept personal criminal exposure across multiple jurisdictions, with no legal entity to shield you and no compliance path that satisfies the statute. Or write a permissioned, KYC-gated privacy tool that regulators can live with, that has a corporate wrapper, and that a lawyer can defend. The second option is safer. It is also less useful. The market is being pushed toward privacy that only exists where it is already permitted — which is to say, privacy for institutions, not for people.

This is the talent migration the sector is not tracking. The developers who would have built the next generation of permissionless privacy infrastructure are being quietly redirected. Some are pivoting to compliant privacy — zero-knowledge KYC, auditable anonymity sets, institutional-grade tooling. Others are leaving crypto entirely. A few are going anonymous, which removes the very accountability that makes open-source development trustworthy. None of these outcomes is good for the ecosystem.

And the migration is not neutral. It reallocates talent toward the projects that can afford legal counsel and corporate structures — which is to say, toward well-funded, centrally governed teams. The permissionless ethos loses its builders to the permissioned one. The protocol that survives will be the one that never needed a founder to be prosecuted, because it had a foundation, a compliance officer, and a jurisdiction of incorporation. That is a quieter, slower centralization, and it is happening right now.

The Ecosystem Ripple: Where the Pressure Lands

A developer conviction does not stay confined to the developer. It propagates through the ecosystem in predictable directions.

Start with the protocols themselves. Tornado Cash's contracts remain live on Ethereum — code cannot be un-deployed. But the front end, the user-facing interface that made the protocol usable, has been shut down. The result is a zombie protocol: technically alive, practically abandoned, used only by those sophisticated enough to interact with the contracts directly. The value that once flowed through it has scattered. This is what regulatory pressure looks like at the application layer — not deletion, but paralysis.

Move to exchanges. Privacy assets — TORN and the broader anonymity set — face escalating delisting pressure at compliant venues. Each legal development in this case gives exchanges another reason to remove trading pairs, citing risk management. Delisting is a one-way ratchet: it is far easier to remove a pair than to add it back. The liquidity that exits does not return quickly. For holders of privacy-sector tokens, the practical effect is a slow bleed of market access, not a dramatic crash.

Move to DeFi. Protocols that previously integrated privacy tools are quietly severing those integrations and hardening their compliance posture. The rational move for any DeFi team is to preemptively distance itself from anything a prosecutor might call a mixer. This reduces the anonymity set available to legitimate users, which reduces the actual privacy the tools provide, which reduces their utility. The regulatory pressure degrades the product even before any ruling is final.

And then there is the sector that benefits. RegTech — on-chain analytics and forensic firms — has its business model validated by every case like this. The government's ability to trace funds through a mixer, to build the evidence that convicted Sterlingov and that anchors the Storm prosecution, depends entirely on commercial blockchain analysis. Every conviction is a testimonial. The firms that sell surveillance are the quiet winners of the privacy wars, and their revenue is a direct function of the enforcement intensity that is rising.

What the Bulls Got Right

I have spent most of this piece dissecting the prosecution's strategy, which is the honest thing to do when the prosecution is winning the procedural battle. But a forensic audit that only counts the losses is a bad audit. The bulls are not wrong about everything, and the places where they are right deserve to be named.

First, the deadlock is real evidence of something. The government threw its full weight at Roman Storm — money laundering, sanctions evasion, money transmitting — and walked away with one count. Two of three collapsed into a hung jury. That is not nothing. It suggests that a jury of ordinary people, presented with the full machinery of federal prosecution, was not convinced that writing open-source code is the same as running a criminal enterprise. The government's theory is not self-evidently true to a lay audience. That matters for retrial strategy, because the second jury will face the same conceptual gap.

Second, the code-is-speech argument has not been defeated. It has simply not been tested at the right level. The idea that publishing software is protected expression under the First Amendment is not fringe; it has serious legal pedigree. The reason it has not carried the day in the Storm case is procedural — it was not the basis of the motion that was denied. But it remains available on appeal, and it is the argument most likely to produce a circuit split and eventually reach higher courts. The bulls who say this is a long game are correct. This case is the opening, not the close.

Third, the compliance-privacy thesis is genuine, not cope. If permissionless privacy is criminalized, the demand for privacy does not disappear. It moves. The projects building zero-knowledge privacy with compliance hooks — the ones that can show a regulator a valid, auditable trail — are positioned to absorb the institutional demand that Tornado Cash used to serve. That is not a betrayal of the ethos; it is a recognition that privacy has to survive somewhere, and it will survive where the law allows it. The bulls who bought compliant privacy exposure before the crowd are, on the evidence, making a defensible bet.

Fourth, and most uncomfortable for the bearish case: the protocol is still running. You can shut down a front end. You can convict a developer. You cannot switch off a deployed contract on a sufficiently decentralized chain without a level of consensus coordination that would itself be a scandal. The technology's persistence is the strongest argument that the code is, in some meaningful sense, separate from its authors. The prosecution needs that separation to fail. The bulls are betting that it holds.

The Forensic Audit Isn't Over

Every blockchain story ends in a forensic audit, and this one is barely past its first ledger entry. What we have is a citation, a deadlock, a scheduled hearing in April 2026, and a money transmitting conviction that is being appealed. None of it is final. All of it is consequential.

The forward-looking judgment is this: the DOJ is not trying to win a single case. It is trying to build a body of authority that makes the next case easier. Bitcoin Fog is the precedent. Tornado Cash is the test. If the Southern District accepts the analogy, the precedent set will not be about mixers. It will be about whether the architecture of a protocol can be legally erased — whether a court can look at a non-custodial contract and see a custodial business, at a developer and see an operator, at a hash and see an office address.

The question that should keep every permissionless developer awake is not whether Roman Storm is guilty. It is this: if jurisdiction follows the code, and the code has no location, then where exactly are you legally standing when you deploy?

The answer the courts give will define who is allowed to build in the open. And the developers who intend to keep building are running out of places to stand.