What a Search Warrant Actually Allows FBI Agents to Take

Last updated 2 weeks ago. Our resources are updated regularly but please keep in mind that links, programs, policies, and contact information do change.

On the morning of February 25, 2026, FBI agents arrived at two locations: the San Pedro home of Alberto Carvalho, superintendent of the Los Angeles Unified School District, and LAUSD’s downtown headquarters. A third team carried out a coordinated search at a property in Southwest Ranches, Broward County, Florida. By the end of the day, agents had seized materials from all three locations. No charges had been filed. The specific legal reason for the investigation remained sealed.

The Fourth Amendment promises that “the right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated.” The Constitution requires that warrants particularly describe the place to be searched and the items to be seized. In theory, this is a strong protection. In practice, it is much weaker than it sounds.

The Carvalho search illustrates how search warrant authority works when federal investigators arrive at a public official’s home and at an institution holding federally protected records for hundreds of thousands of children.

The Particularity Requirement: What It Promises

The Supreme Court stated the particularity requirement’s purpose plainly in the 1927 case Marron v. United States: “The requirement that warrants shall particularly describe the things to be seized makes general searches under them impossible and prevents the seizure of one thing under a warrant describing another. As to what is to be taken, nothing is left to the discretion of the officer executing the warrant.”

That requirement does not mean investigators can only take the exact items named in the warrant. The Supreme Court has recognized a “plain view” exception that allows seizure of items not listed if officers are lawfully present, they observe the item in plain view, and the incriminating nature of the item is immediately apparent. (The plain view doctrine sounds reasonable in theory; it becomes much broader in a digital search.)

An agent searching a computer for financial fraud documents can come across thousands of other files in the same directory. Once the agent is lawfully accessing the system, all of those files are arguably in plain view. What counts as “immediately apparent” incriminating nature leaves a lot of wiggle room in how courts and agents interpret it.

The 2004 Supreme Court decision in Groh v. Ramirez illustrates how the system can fail even when the defect is obvious. ATF agent Jeff Groh prepared a warrant application to search a Montana ranch for firearms. His supporting affidavit described the firearms in detail. The warrant itself, however, did not describe the items to be seized at all. Groh had mistakenly typed a physical description of the ranch in the space reserved for items.

The warrant was defective on its face. The search happened anyway. The Court ultimately held the search unconstitutional and denied Groh qualified immunity, the conditional protection shielding officials from personal liability only when they do not violate clearly established rights a reasonable person would have known. That ruling came after years of litigation. The evidence was excluded, but only because someone fought it all the way to the Supreme Court.

Most people do not have the resources to do that. And even when they do, the structural problem of one-sided warrant review remains a real concern, though the nature of that problem differs significantly depending on the court.

The Foreign Intelligence Surveillance Court, a specialized secret court that handles national security wiretap and search requests, approved 33,942 warrants over a 33-year period with only 12 denials. That figure is frequently cited as evidence that judicial warrant review is a rubber stamp, but it is a poor proxy for criminal warrant practice. FISA uses a probable cause standard, but one requiring a showing that the target is a foreign power or its agent — not that a crime has been or will be committed. That is a meaningfully different threshold from criminal probable cause.

Its judges hear only the government’s submissions with no defense bar present. There is no later real back-and-forth review where someone pushes back on the government’s claims. The institutional incentives are different from those facing a federal magistrate judge reviewing a Rule 41 criminal warrant application.

Criminal magistrates operate in a system where defense counsel can later challenge the warrant in suppression proceedings. This creates reputational and reversal consequences that FISA judges do not face. A magistrate who routinely approves overbroad or poorly supported applications will see those applications challenged, litigated, and sometimes reversed on appeal. That feedback mechanism does not exist in the FISA context.

The concern about one-sided review is not baseless in the criminal context either. The government pre-screens its applications and does not submit cases where prosecutors doubt probable cause exists. The magistrate sees only the government’s presentation. The target is not present.

Whether that process works as a genuine check or a procedural formality is a question the FISA data cannot answer. The criminal warrant literature, while less dramatic, suggests approval rates in criminal matters are also high, reflecting both government self-selection and the relatively low bar of probable cause.

What Happens When the Target Is an Institution, Not a Person

A search warrant for a private home is complex enough. A search warrant for a large public institution is a different problem entirely.

LAUSD headquarters is not a house where one person lives and works. It is a government building where hundreds of employees work across multiple departments. The district’s records include personnel files for thousands of employees, communications with the district’s legal counsel, contracts with vendors, and grant applications. Most importantly, they include educational records for a large number of students (current enrollment stands at approximately 398,487 for 2025-26, though total records held may include former students).

Many of those students come from immigrant families. The district does not formally track or estimate the undocumented student population, as California schools do not collect immigration status data; available data focuses on newcomer students, who comprised approximately 4.4% of enrollment in 2025-26. That last fact matters: a seizure of student records from this institution is not a distant privacy concern.

Those student records are protected by federal law. The Family Educational Rights and Privacy Act of 1974 establishes that schools receiving federal funding must protect the confidentiality of education records. The statute defines these as records directly related to a student and maintained by an educational institution. FERPA does contain a law enforcement exception: schools may share records under a lawful court order or subpoena. A federal search warrant is a court order. But whether a search warrant automatically satisfies FERPA’s specific procedures for law enforcement access is genuinely unsettled.

The Department of Education’s guidance says schools “may typically disclose information from student records without consent only under a FERPA exception or in response to a valid court order or subpoena” and that schools “should consult with legal counsel to determine whether the request is valid and appropriately tailored.” The key word is “may.” Schools have discretion. But when federal agents arrive with a warrant, schools rarely use that discretion by pushing back.

The practical assertion of FERPA rights requires having counsel on site who knows the statute and is willing to raise objections in real time. That means raising objections while agents are standing in the building. That is a difficult position to be in.

The California Attorney General issued updated guidance in December 2024 helping K-12 schools understand the distinction between judicial and administrative warrants in the context of immigration enforcement. But that guidance addresses immigration enforcement specifically. The broader question of how FERPA interacts with a general federal criminal search warrant remains unresolved. The students whose records sit in LAUSD’s servers have no way to contest the seizure before it happens.

The AllHere Investigation: What the Government Appears to Have Been Looking For

The underlying facts behind the search appear to center on LAUSD’s contract with AllHere, an AI company that signed an approximately $6.2 million agreement with the district in June 2023 to develop an AI chatbot called “Ed.” The company’s collapse was rapid and documented. By June 2024, AllHere had furloughed most of its employees. By July 2024, it had filed for Chapter 7 bankruptcy liquidation.

In November 2024, AllHere’s CEO, Joanna Smith-Griffin, was arrested and charged with multiple counts of fraud. According to the indictment, Smith-Griffin told potential investors that AllHere generated $3.7 million in revenue in 2020 with $2.5 million in cash. The company had in fact generated approximately $11,000 in revenue. She also allegedly falsely claimed contracts with the New York City Department of Education and with Atlanta Public Schools.

A whistleblower, Chris Whiteley, AllHere’s former senior director of software engineering, raised separate concerns: that student records were stored in overseas databases in violation of LAUSD’s data privacy policies. That allegation, if confirmed, would implicate federal law and potentially federal funding compliance.

The simultaneous search of a Florida property adds another layer. That property is associated with Debra Kerr, a consultant with long ties to Carvalho from his time as superintendent in Miami-Dade. Reports suggest Kerr worked as a consultant to AllHere and has claimed the company owes her significant unpaid commissions, though the specific details of her role in the investigation have not been confirmed in public records.

For the FBI to search that property, investigators needed a separate warrant application approved by a magistrate judge in the Southern District of Florida. As a general rule, federal search authority does not carry over automatically across district lines — each location requires its own warrant, probable cause showing, and judicial approval, though Rule 41(b) carves out exceptions for property likely to move before execution and for terrorism-related investigations.

That multi-district requirement is a formal structural check. Whether it provides meaningful added protection is another question. The magistrate judges in Los Angeles and Florida would both be reviewing the same underlying facts. If the government’s account of the facts persuades one judge, it will almost persuade the other.

What Agents Can and Cannot Lawfully Seize

The table below summarizes the legal rules governing what FBI agents can take during a search, alongside the practical reality of what happens when those rules are tested. The gap between the two columns is where most of the real action is.

Sources: Federal Rule of Criminal Procedure 41(g) allows “a person aggrieved by an unlawful search and seizure of property or by the deprivation of property” to move for the property’s return. (Note: the rule uses only the term “aggrieved” without internal definition.) That sounds like a real remedy. In practice, it is far more limited.

To win a Rule 41(g) motion, the person filing the motion must prove both that they have a right to the property and that the seizure was unlawful. The government can defeat the motion on multiple grounds — showing the property is still needed as evidence, that the claimant lacks lawful possession rights, or that the property is contraband subject to forfeiture. There is no statutory deadline by which the government must charge someone, though under 18 U.S.C. § 983(a)(3)(A), once a claim is filed the government has 90 days to file a civil or criminal forfeiture complaint, return the property, or seek a court extension for good cause. An investigation can last months or years, and prosecutors can hold seized materials for extended periods by continuing to describe them as potentially needed for a future prosecution.

The good faith exception, established in United States v. Leon, makes this even harder. It is a rule that says evidence can still be used if agents genuinely believed the warrant was valid, even if it later turns out it was not. Even if a warrant is later found defective, evidence may still be admissible if investigators reasonably relied on the warrant at the time of the search. This exception makes it much harder to get evidence thrown out by challenging the warrant as a remedy for overbroad seizures.

Reports suggest that federal suppression motions based on warrant defects succeed in a small fraction of cases. The structural reasons are not hard to see. The target lacks access to the warrant application, must actively prove the search was unlawful, and faces a government that can cite qualified immunity (the conditional legal protection that shields individual agents from personal liability only when they did not violate clearly established statutory or constitutional rights of which a reasonable person would have known) for individual agents.

There is also a standing problem specific to institutional targets. LAUSD as an institution is not a criminal defendant. It cannot be charged with a crime. But its records have been seized. The district has a practical interest in accessing those records.

Yet institutions typically don’t have the legal right to bring this kind of challenge by filing a Rule 41(g) motion, which requires that the person filing be “aggrieved by an unlawful search and seizure.” When the government searches a company’s offices and seizes its files, it is genuinely unclear who can bring the motion: the company? Individual employees? Both? The statute does not clearly answer. As a result, institutional searches often proceed with limited meaningful challenge.

For a deeper look at how sealed warrants interact with the challenge process, our earlier analysis of sealed warrants and election records covers the transparency problems in detail.

The Taint Team Problem

When FBI agents search an office, they will almost certainly encounter documents protected by attorney-client privilege. An email from Carvalho to LAUSD’s general counsel seeking legal advice about the AllHere contract would be privileged. Communications between the district’s legal staff and the superintendent’s office discussing potential legal risk from the AI chatbot project would also be privileged. These materials cannot lawfully be used in a prosecution.

To protect privilege while allowing investigations to continue, the Department of Justice uses what it calls a “taint team” or “filter team”: an internal group of DOJ attorneys and staff kept separate from the investigation. After the search, the filter team reviews seized materials to separate privileged documents from non-privileged ones. Only materials deemed non-privileged are passed to the investigation team.

In practice, this has a built-in flaw that courts have repeatedly pointed out: the filter team operates within the government. The target has limited ability to challenge the filter team’s privilege determinations. Legal scholars and defense practitioners have argued that the appearance of independence is weakened by the government’s obvious interest in maximizing the evidence available to prosecutors. Critics of DOJ privilege teams have noted that if a filter team member wrongly identifies privileged material as non-privileged and prosecutors see it, the damage may be irreparable. You cannot un-read a document.

For Carvalho, this is especially sharp. The communications most likely to be sought by prosecutors investigating the AllHere contract are precisely the ones most likely to be privileged. What legal advice did he receive about potential risks? What did LAUSD’s attorneys tell him about the contract’s terms? The filter team is supposed to prevent prosecutors from seeing those communications. Whether it does is a question that cannot be answered from the outside.

The Political Dimension: What the Law Does and Does Not Protect

Carvalho has been a nationally well-known critic of the Trump administration’s immigration enforcement policies around schools. In September 2025, California Governor Gavin Newsom signed legislation designed to shield schools from ICE officers — an event Carvalho was reported to have attended, though his presence at the signing ceremony has not been independently confirmed. He has stated he would “fight for those who find themselves in the same predicament I faced over 40 years ago” as a formerly undocumented immigrant himself. The search occurred weeks into the administration’s second term, as immigration enforcement intensified across the country.

The obvious question is whether the investigation is politically motivated. Before examining what the legal framework does and does not protect against, it is worth looking carefully at the factual record on that question: because the timeline significantly complicates the political motivation hypothesis.

The AllHere fraud charges were not started by the current administration. Joanna Smith-Griffin was arrested in November 2024 and charged by career federal prosecutors operating under the prior administration. The whistleblower allegations about overseas student data storage represent a potential federal funding compliance violation. They similarly predate any change in administration and were developed independently of the political context surrounding Carvalho’s immigration advocacy.

The multi-district warrant requirement adds a further structural consideration. For investigators to search the Florida property associated with a consultant connected to the AllHere contract, prosecutors needed to persuade an independent magistrate judge in the federal court covering South Florida to approve a separate warrant. That judge reviewed the same underlying facts.

Two independent judicial officers in two jurisdictions reviewed the government’s probable cause showing. That is not a guarantee against abuse, but it is a meaningful structural constraint that makes a purely retaliatory investigation harder to carry out without a genuine factual foundation.

The strongest version of the government’s position, then, is not merely that career investigators were involved. It is that the core factual basis for this investigation was assembled and acted upon before the political context that makes the timing appear suspicious. A careful skeptic of the political motivation theory would note that the AllHere fraud was independently charged and the whistleblower’s concerns were documented. The investigation’s expansion to Carvalho may reflect the ordinary progression of a fraud case toward the public officials who approved the contracts. It may not reflect a decision made in response to his immigration advocacy.

That said, the legal framework does not provide a clean mechanism for separating legitimate investigations from politically motivated ones, and the concern is not trivial. The particularity requirement does not prohibit searches of political opponents. The Fourth Amendment does not require a magistrate judge to ask whether a search is politically motivated. The exclusionary rule targets procedural constitutional violations, not investigative motive. A separate First Amendment retaliation claim could theoretically provide independent relief, but it requires proving both retaliatory intent and the absence of any legitimate law enforcement basis. A perfectly particularized warrant for a vocal critic of the administration is constitutionally valid even if the investigation was started in retaliation for protected speech. Protected speech means speech the government cannot legally punish.

The First Amendment might provide some protection against investigations motivated purely by retaliation for protected speech. But that protection requires proving that the investigation was motivated by protected speech and that no legitimate law enforcement basis existed. When there is a documented factual basis, as there is here, given the independently charged AllHere fraud and the whistleblower’s allegations, that showing becomes difficult to make. The existence of a legitimate basis does not rule out the possibility of mixed motives. It does mean, however, that political motivation, even if present, would not by itself render the search unlawful.

This is the same tension we examined in our coverage of the legal standard for searching a journalist’s home: a federal investigation can have both a legitimate law enforcement basis and a political dimension at the same time. The law does not have a clean mechanism for separating the two. The magistrate judge sees only the government’s presentation of facts. The target cannot contest the warrant before execution. Once the search has happened, the burden of proving improper motivation falls on the person who was searched.

Career investigators, not political appointees, make day-to-day decisions about whether to pursue cases. The internal FBI approval process for significant warrant applications involves multiple layers of review before an application reaches a magistrate judge. In this case, those institutional checks operated against a backdrop of independently developed fraud charges and whistleblower allegations. That factual foundation predates the political context. Independent judicial officers in two jurisdictions found it sufficient to support probable cause.

Those checks are real, and in this instance they are reinforced by the timeline. They are also, however, internal to the government and not subject to adversarial scrutiny. Whether the investigation’s scope and timing were driven entirely by the evidence, or whether political considerations shaped prosecutorial priorities, cannot be determined from the public record.

What Carvalho and LAUSD Can Do Now

The morning after a federal search, the options are real but limited, and mostly reactive.

The first step is demanding a full inventory of all items seized. Under federal rules (Rule 41(f)), executing officers must provide an inventory of seized property. Comparing that inventory to the warrant’s scope is the only way to identify items that may have been taken outside the warrant’s authorization.

The second step is hiring criminal defense counsel immediately, if not already done. Carvalho should assume potential charges and prepare accordingly. LAUSD should consider hiring outside counsel specializing in federal criminal defense and education law, separate from its regular legal staff. The district’s own attorneys may face attorney-client privilege complications. Their own communications with Carvalho may be part of what was seized.

A Rule 41(g) motion for return of property is worth considering, particularly for items central to the district’s operations or items that appear to fall outside the warrant’s scope. That motion will be difficult to win while the investigation is ongoing. But it may result in partial return of property or, at minimum, access to forensic copies of digital evidence so that LAUSD can keep functioning.

Requesting unsealing of the warrant application is also worth attempting. Courts are generally reluctant to unseal warrant materials during active investigations — a distinct question from the low success rate of Rule 41(g) property return motions — but the justification for continued sealing weakens once the search has been carried out and the investigation’s existence is public knowledge. Unsealing would allow Carvalho and LAUSD to understand what probable cause the government relied on. It would also allow them to identify factual claims that may be disputed and to determine whether the warrant’s scope was adequate or overbroad. As we have previously covered in our analysis of federal seizure authority and state records, sealed warrants create a transparency problem that persists long after the search itself.

For LAUSD specifically, there is a FERPA angle worth pursuing. If investigators seized student records without complying with FERPA’s specific procedures for law enforcement access, the district may have grounds to challenge the seizure of those records specifically and to seek Department of Education enforcement action. FERPA violations do not automatically suppress evidence in a criminal case. But they can create separate legal problems for the government, such as Department of Education enforcement action, and may give the district use in negotiations over the return of district records.

The Unresolved Question

Here is what remains genuinely open: whether the magistrate judge who approved these warrants imposed meaningful limits on the search’s scope, or whether the warrant language was broad enough to give investigators wide discretion in deciding what to take from a school district’s headquarters. That question cannot be answered until the warrants are unsealed. The warrants may remain sealed for months or years, or indefinitely if no charges are filed.

This is the part of the Fourth Amendment framework that gets the least public attention. The Constitution’s protections against unreasonable searches are real. The particularity requirement is real. The magistrate judge’s role is real. But the enforcement mechanisms for those protections are almost entirely after the fact. You find out what the government took after they take it. You challenge the seizure after the fact and bear the burden of proving the search was unlawful. You do all of this without access to the document that would tell you what the government told the judge to obtain permission to search in the first place.

The AllHere investigation may ultimately produce charges that are entirely justified by the evidence. Or it may close without charges, leaving Carvalho with a search on his record and no clear mechanism for public vindication. What this case has already shown, regardless of how it ends, is that the gap between what the Fourth Amendment promises and what it delivers in practice is widest precisely when the stakes are highest. That is when the target is a public official, when the institution holds sensitive records about vulnerable people, and when the investigation’s motivations cannot be publicly scrutinized because the warrant remains sealed.

The one-sided warrant approval process, the weak standing rules for institutional targets, and the inadequacy of filter team procedures have been documented in detail by legal scholars, the Fourth Amendment Center, and federal public defenders for decades — in law review articles, suppression litigation, and other venues. These critiques are not abstract; they have been made with detail and evidence in forums where they could have prompted reform.

This case does not make these problems visible for the first time. What it adds is a high-profile institutional dimension: the combination of FERPA-protected records for a large number of students, a multi-district warrant structure, and a publicly well-known target. That combination is genuinely new and may draw legislative or judicial attention that prior cases did not.

Whether that attention produces reform, and whether the investigation itself ultimately proves justified, are separate questions. The public record cannot yet answer either of them.

Our articles make government information more accessible. Please consult a qualified professional for financial, legal, or health advice specific to your circumstances.