Most lawyers who followed United States v. Nejad recall it as a severe prosecutorial misconduct case. A jury convicted Ali Sadr Hashemi Nejad on five counts in March 2020; within months the government itself asked the court to undo the verdict, and Judge Alison Nathan vacated the conviction, granted a new trial, and dismissed the indictment with prejudice. Her September 2020 opinion is a catalogue of disclosure failures — exculpatory evidence turned over late or not at all, prosecutors caught on their own internal messages proposing to “bury” a document among already-disclosed exhibits, and a misrepresentation to the court about how that document had been disclosed. It is required reading on Brady and candor, and that is how it’s remembered.
But there is a second important issue inside that case, and it’s the one this blog cares about. Folded into the same opinion is a digital-search revelation that drew far less attention and matters at least as much for the questions I’ve been building all week — because it is a textbook instance of the analysis-scope problem, admitted by the government, in a real prosecution, with real consequences.
Here is the digital piece, drawn exclusively from the public opinion. Before the federal case existed, the Manhattan District Attorney’s office investigated the matter as a state crime and obtained search warrants for a set of email accounts, including the defendant’s personal email. Those were state warrants: the supporting affidavit invoked ‘reasonable cause’ to believe the accounts held evidence of New York money-laundering offenses, and the warrant authorized the District Attorney’s office to search the returns for that state-law material. Some of those emails eventually reached the U.S. Attorney’s office, which found them, in its words, particularly incriminating. The defendant moved to suppress; the court allowed most of the evidence in.
Now the part pertinent here. Throughout the suppression fight, the government consistently represented that the state warrant returns had been searched by the District Attorney’s personnel for material pertinent to the state-law crimes the warrants named — that state reviewers went through the returns, and only then passed “hot docs” up to the federal prosecutors. Years later, after trial and after conviction, the government admitted that this was not what happened. As the September opinion recounts, federal investigators had in fact been mining the state warrant returns for federal crimes, without a warrant of their own — the FBI had state personnel search the email data in support of a federal investigation and federal interviews, in the opinion’s words “contrary to arguments [the government] made during suppression litigation.” The court’s assessment was blunt: this conduct was “likely unconstitutional,” because — quoting the Second Circuit’s settled rule — “a search must be confined to the terms and limitations of the warrant authorizing it.”
That was September. Five months later it turned out to be far worse than “a few interviews.” In a February 2021 opinion, the court disclosed what the government had finally confessed: this was not a handful of ad hoc requests. The FBI had taken the entirety of the raw email search-warrant returns from the District Attorney, loaded them into a proprietary FBI document-review database — the opinion names it BIDMAS — and then run approximately 636 searches of that database over roughly two years, hunting for material relevant to its federal investigation, under state warrants that authorized searching only for state-law crimes. The court again called it likely unconstitutional under the same rule: those queries could not have conformed to warrants that never authorized them. And it added the detail that turns a violation into an indictment of a system — the misuse was “much more systematic” than first admitted; the episode “raises questions about the FBI’s attention to the limits of the Fourth Amendment in its use of the BIDMAS system more broadly”; and the court ordered the government to go find the other cases in which the same thing had happened, and to notify those courts and defendants.
Sit with those numbers, because they are the whole argument in miniature. One warrant authorizes taking a body of email for one purpose. The government copies all of it into a searchable database and queries it six hundred–plus times for a different purpose entirely. No court authorized the queries, no rule bounded them, no record surfaced them for years — and the only reason we know the count at all is that a wholly separate scandal forced the government to reconstruct its own conduct under a series of court orders. That is not a wiretap-era problem. That is the analysis stage itself — a database, a query box, and an intrusion that leaves no footprint — operating with no law addressed to it.
Read that against Tuesday’s post and it should look familiar, because it is the § 5 problem in the wild. The data was lawfully collected — no one said the state warrants were invalid. The wrong was in the analysis: examining lawfully-seized returns for a purpose the warrant that authorized seizing them never contemplated. A warrant to search email for state money laundering is not a license to search the same email for federal sanctions crimes, any more than — to reuse Tuesday’s line — a warrant to search a house for a stolen car authorizes reading the homeowner’s diaries. Collection was lawful; the analysis exceeded it; and the Fourth Amendment violation, if there was one, lived entirely at the analysis stage — the stage the law does not regulate, and the one § 5 exists to reach.
Three things about how this surfaced are worth sitting with, because they are the real lesson.
First, lawful collection did nothing to prevent an unlawful analysis. Valid warrants were in hand, and they were no obstacle at all to searching the returns for the wrong crime. This is the entire argument for a § 5: the warrant at the front end is not a substitute for rules governing what happens to the data after it’s seized. A regime that polices only collection would have graded this search a pass.
Second — and this is the part that should unsettle you — the violation was invisible for years, and while it was invisible it was doing work. The government did not merely fail to mention the database. During the suppression fight it affirmatively represented that it had searched only the “hot docs” the District Attorney had already flagged as responsive to the state warrants, and the court relied on that representation in deciding the disputed suppression issues. So the hidden analysis-stage violation was not a dormant secret — it was the false premise underneath a ruling that let the evidence in. And it surfaced only because the prosecution had already collapsed for unrelated reasons, forcing the government to reconstruct its own conduct under a series of court orders. Absent that collapse, by all appearances no one outside the government would ever have learned that the returns had been loaded into a database and searched six hundred times. Scope violations at the analysis stage are invisible from the outside by their nature, because there is no record of what was searched and for what — the intrusion happens inside an office, in a database, on a set of returns, unseen. Without a mandatory record of the analysis — the handling record this blog keeps circling — the only way such a violation comes out is by accident.
Third, and connecting back to a theme from earlier this week, the line got drawn and the remedy evaporated. Both opinions flagged the conduct as likely unconstitutional; neither adjudicated it, because the indictment had already been dismissed and, as the court put it, there would be “no further litigation of these issues before the Court.” The suppression question the defendant had raised — the one the government now conceded it had argued on a false premise the court had relied on — would never be resolved on the merits. And the response, in the end, was referral: to the Department of Justice’s Office of Professional Responsibility, and to the FBI’s own Inspection Division to determine whether the agents running the database searches even understood that they exceeded the warrants, together with an order for the government to go identify the other affected cases itself. Serious enough that the government cited the suppression risk as a reason to dismiss the untried co-defendant’s case; serious enough to refer to two internal investigative bodies and to demand a hunt for other victims; and still never ruled on. If that shape sounds familiar, it’s the same one the retention cases keep tracing: the digital Fourth Amendment question that everyone can see and no court quite reaches.
Let me be fair about the limits of what the opinions establish, because I’m holding this to their four corners. The court said “likely” unconstitutional, not that it was; the constitutional question was never litigated to judgment, precisely because the case was dismissed. Parallel state and federal investigations are genuinely common and can be messy, and reasonable lawyers can argue about how much cross-use of lawfully-obtained returns the Fourth Amendment tolerates. And the headline wrong in Nejad was the disclosure misconduct, not the search — in the opinions’ own structure, the search issues sit in a section of their own, apart from the Brady drama that dominated the coverage. None of that is a reason to look away. If anything, it’s the reason to look harder: a scope violation this systematic — six hundred database queries across two years — this consequential, and this close to un-litigated would have vanished without a trace but for a separate scandal. The routine version, the one with no collapse to shake it loose, is happening where no opinion will ever describe it.
That’s why the model act treats analysis as its own regulated stage, and why it pairs scope rules with a handling record and a duty of candor about what was searched and what items were properly seized. Nejad shows all three failing at once: the search exceeded the warrant’s scope, no record made the excess visible, and the government’s account of what had been searched turned out to be contrary to the facts. Fix any one of those and the violation surfaces; fix all three and it mostly stops happening. The Fourth Amendment’s future won’t be secured at the courthouse door where the warrant issues. It will be secured — or lost — after the seizure, in the review of the returns, in the part of the process Nejad let us see only because everything else had already gone wrong.
Next Tuesday, § 6: copying and retention — how the government comes to hold these returns and images in the first place, and how long it gets to keep them.
Leave a comment