What if making twenty-five million dollars in twelve seconds on the Ethereum network did not actually break the law? That is the question a federal judge just answered in a ruling that left a lot of people in crypto scratching their heads and others quietly celebrating. Two brothers with MIT educations walked out of court cleared of serious criminal charges, and the decision turns on something surprisingly basic: whether their actions counted as deception in a system that was never designed to be fair or polite.
The Twelve-Second Trade That Changed Everything
I still find it hard to wrap my head around the speed. Twelve seconds. That is all it took for Anton and James Peraire-Bueno to extract roughly twenty-five million dollars from a series of trades that targeted sandwich bots on Ethereum. Prosecutors called it fraud. The judge called it something else entirely. In a detailed forty-six-page opinion, the court concluded that the government simply could not prove the brothers made any material misrepresentation. Without that key element, the wire fraud charges and the related conspiracy counts collapsed.
This was not some obscure technical dispute that only blockchain developers care about. The case sat at the intersection of code, markets, and criminal law. It forced a federal judge to stare directly at the messy reality of how Ethereum actually works and decide whether the usual rules of traditional finance still apply when the entire environment is openly adversarial by design.
Understanding the Environment the Brothers Operated In
Ethereum is not a polite auction house. It is a public, permissionless system where participants constantly try to extract value from the order of transactions. Maximal extractable value, or MEV, has become one of the most studied and most controversial parts of the network. Sandwich trading sits right in the middle of that world. A sandwich bot sees a pending buy order, buys the same token first, lets the original transaction push the price up, then sells into the higher price. The original trader ends up with worse execution. The bot walks away with the difference.
The brothers spent months building something they called Omakase. They registered sixteen validators. They studied the way certain relays handled pending transactions. Then, in April 2023, they executed a carefully prepared sequence. Eight of their transactions attracted sandwich bots. Instead of letting those bots complete their expected sales, the brothers rearranged the block in a way that left the bots holding the bag. The profits landed almost immediately. Three sandwich traders took losses. One of them, according to court records, was a firm called Savannah Technologies.
Here is where things get interesting. The brothers took advantage of a temporary vulnerability in the relay software connected to MEV-Boost, an open-source tool that many validators use. That vulnerability allowed them to see transaction data they would not normally see until after they had committed to a block. They used that information. They altered the proposed arrangement of transactions. And they made a great deal of money in a very short window.
It is a trustless, highly adversarial environment with no terms of use.
That single sentence from the judge captures the heart of the ruling. Ethereum does not come with a customer service agreement. There is no user handbook that promises your transactions will be treated fairly or executed in a particular order. Participants submit transactions to a public mempool knowing full well that others may try to profit from them. The court refused to invent obligations that the protocol itself never created.
Why the Spoofing Comparison Fell Apart
Prosecutors tried to compare the brothers’ strategy to spoofing in traditional markets. Spoofing usually involves placing large orders with the clear intention of canceling them before they execute, simply to create a false impression of supply or demand. The judge found the comparison unpersuasive. The brothers’ so-called bait transactions were never withdrawn. They appeared on the blockchain within about a minute. They were real transactions that settled, not phantom orders designed to mislead and then disappear.
I have watched enough market manipulation cases over the years to know that the line between aggressive trading and outright fraud can feel thin. Yet in this instance the court drew a fairly clear distinction. Submitting a transaction to the public pool does not equal a promise that the transaction will sit quietly next to other trades in some predetermined sequence. The network simply does not work that way, and never has.
Another part of the government’s case focused on a digitally signed block header that contained incorrect information. The judge acknowledged the information was wrong. But the evidence showed the relay in question never actually examined those particular fields before releasing the transaction data. Because the false information had no practical effect on the recipient’s behavior, it could not support a fraud conviction. That is a high bar, and the prosecution did not clear it.
The Long Road From Arrest to Acquittal
Authorities arrested the brothers in May 2024 after an investigation that involved the Internal Revenue Service’s Criminal Investigation unit. Each of the three original charges carried a potential twenty-year maximum sentence. That is not a light cloud to live under. In July 2025 the same judge denied an earlier motion to dismiss the fraud charges, though a separate count involving conspiracy to receive stolen property was dropped when prosecutors chose not to pursue it.
The trial itself ended in a mistrial in November 2025 after the jury could not reach a unanimous verdict. That outcome already suggested the case was far from straightforward. Then came the November 2026 ruling that effectively ended the criminal prosecution. The money-laundering conspiracy charge depended on the existence of fraud proceeds. Once the fraud theory failed, that charge failed with it.
Flashbots, the group behind the MEV-Boost software, reacted quickly at the time of the original incident. They temporarily blacklisted the validator involved and patched the relay software within twenty-four hours. A wallet holding about three million dollars of the proceeds was frozen. By October 2023 the brothers had already converted the rest of their profits into U.S. dollars. The legal fight, however, stretched on for years.
What This Ruling Actually Means for Crypto Traders
Perhaps the most interesting aspect is how carefully the court treated the nature of Ethereum itself. The judge examined whether participants make explicit or implicit promises simply by submitting transactions, registering validators, or interacting with the software that organizes blocks. The answer, at least in this case, was no. Expectations held by other users about how a platform “should” work do not automatically create a legal duty for everyone else.
That logic echoes another recent wire-fraud decision involving a trader on a different crypto platform. There, too, the court declined to treat other users’ assumptions as binding promises made by the defendant. These rulings, taken together, suggest that federal judges are becoming more reluctant to stretch traditional fraud statutes across every aggressive strategy that appears on a blockchain.
Does that mean anything goes? Of course not. Outright theft, phishing, and clear misrepresentations of fact remain illegal. But the pure exploitation of publicly available information and temporary software quirks, at least when no material lie is told to a specific counterparty, appears harder to prosecute as wire fraud under the standards applied here.
I keep coming back to the speed of the original operation. Twelve seconds of carefully prepared activity generated a sum that most people will never see in a lifetime. The fact that it was possible at all says something important about the current state of on-chain markets. Liquidity is deep enough, bots are sophisticated enough, and the tooling around block construction is complex enough that large sums can move in the blink of an eye. That reality is not going away.
The Role of Validators and Relays in the Drama
Validators sit at the center of Ethereum’s proof-of-stake system. They propose blocks and earn rewards for honest participation. The brothers registered sixteen of them specifically for this operation. When the strategy executed, one of their validators received a small penalty of one ETH for proposing two blocks in the same slot. That detail is almost comical against the size of the overall profit, yet it shows that even successful MEV strategies can carry minor protocol-level costs.
Relays act as intermediaries between block builders and validators. They are supposed to protect certain transaction details until a validator has committed. The vulnerability the brothers used temporarily broke that protection. Within a day the software was updated. The episode serves as a reminder that open-source tooling, while powerful, can contain edge cases that sophisticated actors will eventually find.
In my view, the speed of the patch is one of the more reassuring parts of the story. The ecosystem reacted. The specific attack vector was closed. Yet the broader category of MEV strategies continues to evolve. Every time a new opportunity appears, someone will try to capture it. Courts will then be asked, sometimes years later, to decide whether that capture crossed a legal line.
Lessons From the Evidence Presented at Trial
One detail that stands out is how thin the government’s expert presentation apparently was. Prosecutors did not call an expert witness to explain Ethereum’s trading environment in depth. They also did not put a representative from the Ultrasound Relay on the stand. Only one of the three sandwich traders who lost money actually testified. That is a light evidentiary record for a case of this complexity and dollar amount.
Judges notice these things. When the technical facts are dense and the legal theory depends on proving a specific mental state and a specific misrepresentation, the absence of clear expert testimony can prove fatal. The court essentially said the government had not carried its burden on the element of deception. In criminal cases, that burden is high, and it should be.
I have found that many people outside crypto assume every large, sudden profit on a blockchain must involve some form of cheating. The reality is more nuanced. Some strategies are pure information advantages. Some exploit temporary bugs. Some sit in gray areas that honest people can disagree about. Treating every successful MEV play as fraud would require rewriting the basic architecture of these networks, something no single court is equipped to do.
How Sandwich Trading Fits Into the Larger MEV Picture
Sandwich attacks are unpopular for good reason. Ordinary users hate seeing their trades get worse execution because a bot jumped in front of them. Yet the practice has existed for years and remains profitable for those who run the bots. The brothers essentially ran a counter-strategy against the sandwichers. They became the predator of the predators. Whether that feels morally satisfying depends on your perspective, but the legal question is different from the moral one.
MEV itself is not going to disappear. As long as transactions sit in a public mempool and block proposers have discretion over ordering, someone will try to extract value from that discretion. Research groups, protocol developers, and infrastructure providers continue to work on ways to reduce harmful MEV or redistribute it more fairly. Progress is real but incomplete. In the meantime, aggressive strategies will keep appearing, and some of them will end up in courtrooms.
The brothers’ operation was unusually large and unusually fast. Most MEV extraction happens in smaller increments across thousands of blocks. That difference in scale is probably why this particular episode attracted federal attention. Twenty-five million dollars concentrates minds. Yet the legal principles the court applied do not turn on the size of the profit. They turn on the presence or absence of deception.
Looking Ahead: What Traders and Developers Should Watch
This ruling does not give anyone a free pass to lie or steal. It does, however, reinforce that courts will look closely at the actual rules of the system in question rather than importing assumptions from traditional finance. If a protocol is designed to be adversarial and contains no terms of use that create specific promises, proving material misrepresentation becomes significantly harder.
Developers who build relays, builders who construct blocks, and validators who propose them should still treat security vulnerabilities with urgency. The fact that a strategy is not criminal does not mean it is welcome. Blacklisting, software patches, and social pressure remain powerful tools inside the ecosystem. The brothers’ validator was blacklisted quickly. That kind of response can deter future attempts even when the criminal law stays on the sidelines.
For ordinary traders the practical takeaway is simpler. Slippage and sandwich risk remain real. Using private transaction channels, setting tight slippage tolerances, and understanding how your preferred decentralized exchange handles ordering can reduce exposure. No court ruling will eliminate those risks overnight.
I keep thinking about the human side of the story as well. Two relatively young men with strong technical educations spent years under the shadow of potential decades in prison. The process itself is a form of punishment. When a case ends in full acquittal after a mistrial, it raises questions about charging decisions and about how well prosecutors understand the technology they are trying to regulate through criminal law.
The Broader Tension Between Code and Criminal Law
Crypto keeps forcing traditional legal systems to confront environments that do not match the assumptions built into older statutes. Wire fraud was written for a world of telephones, mail, and later the early internet. Applying it to a global, pseudonymous, adversarial blockchain requires careful translation. Some cases will involve clear lies told to specific victims. Others will involve complex interactions with software that no single person fully controls. Drawing the line correctly matters.
In this instance the court drew the line in favor of the defendants. It refused to treat the mere act of submitting transactions or interacting with public software as an implicit promise that those transactions would behave in a particular way. That refusal feels consistent with how Ethereum actually functions. Whether every future court will reach the same conclusion remains to be seen, but the reasoning here is detailed and careful.
The twenty-five million dollars is already history. Some of it was frozen, some of it was converted to dollars years ago. The legal precedent is what will linger. For anyone who builds, trades, or simply watches this space, the message is worth absorbing. In a system with no terms of use and no central authority promising fairness, aggressive use of public information and temporary software gaps may look ugly, but it does not automatically equal criminal fraud.
That conclusion will not satisfy everyone. Sandwich victims lost money. Critics of MEV will continue to argue that the entire practice extracts value from ordinary users. Those arguments belong in protocol design discussions, research papers, and market competition. They do not automatically belong in a criminal indictment. The judge’s ruling, for better or worse, keeps that distinction intact.
The story of the MIT brothers and their twelve-second operation is not really about two individuals anymore. It is about the boundary between clever code and criminal deception in a network that was built to be trustless. That boundary just got a little clearer. How the industry and future courts respond to the new clarity will shape the next chapter of on-chain markets.
For now, the brothers are free of the charges that once threatened to define the rest of their lives. The software has been patched. The legal theory that tried to criminalize their strategy has been rejected. And the rest of us are left with a useful reminder: in Ethereum, as in many open systems, the absence of rules is itself a kind of rule. Understanding that fact may be the most practical lesson of all.