This is Part 3 of the Digital Dragnets series: Silent Witnesses: When Your Devices Testify Against You. A 4-part investigation into how digital surveillance technology outruns constitutional doctrine, and how courts, corporations, and legislatures are responding to the Fourth Amendment challenges of location data surveillance.

—

TL;DR

While federal appeals courts are locked in a bitter dispute over geofence warrants, several state trial courts have reached a different conclusion about reverse keyword warrants—they can be constitutional when properly particularized. The question of whether the Fourth Amendment protects data you voluntarily share with tech companies is being answered differently depending on whether you’re in federal or state court. The result is geographic asymmetry: what’s protected in a federal prosecution may be admissible in a state court.

—

How Reverse Keyword Warrants Work

Before examining the court rulings, it’s important to understand how reverse keyword warrants function technically.

A traditional search warrant is like searching a specific house: police identify a location (a suspects’s home), obtain a warrant, and search that location for evidence. A reverse keyword warrant works in reverse: police start with the search and then find the person who conducted it.

The technical process:

1. Warrant Request: Law enforcement obtains a warrant compelling a search engine (typically Google) to run a database query.

2. Database Query: The search engine accesses its query database—the logs of every search conducted on the platform. This is not a list of user accounts, but a database of search terms paired with time stamps and, where available, account identifiers.

3. Filtering: The engine filters this database for searches matching specific terms (e.g., “5530 W. 29th Ave.”) within a specified time window.

4. Results: The engine returns identifying information for accounts that conducted the matching searches—typically email addresses, IP addresses, device identifiers, and timestamps.

Key technical distinctions:

– Query Database vs. Account Database: Reverse keyword warrants target the query database (what people searched for), not the account database (who people are). This distinction matters for scale: Google records billions of searches daily across its international user base.

– Logged-in vs. Non-logged-in Searches: Searches performed by logged-in users are linked to specific accounts. Searches by non-logged-in users are linked only to IP addresses and device signatures, which may or may not correspond to Google accounts.

– Temporal Filtering: Most reverse keyword warrants include time constraints (e.g., “searches in the 15 days before the crime”). The database query applies these filters before returning results.

– Geographic Scope: While the warrant may be issued by a state court, the database query potentially accesses search history across Google’s global operations, including international users who searched the target terms.

The controversy arises at step 2: the warrant authorizes searching everyone’s search history—potentially billions of queries—to identify a few suspects. The question courts grapple with is whether this structure violates the Fourth Amendment’s prohibition on general warrants, regardless of how narrow the search term or temporal window may be.

—

The Denver Fire Case

On August 5, 2020, a fire destroyed two connected homes at 5530 W. 29th Avenue in Denver, killing five immigrants from Senegal: Badi, 23; Djibril, 25; Adja, 22; Hassan, 25; and Ismaila, 26—all of whom had recently immigrated to Colorado building new lives in America. The fire, which authorities later determined was intentionally set, had no obvious suspects. Security footage was unavailable. Neighbors reported nothing suspicious. Police had what investigators call a “cold case from day one”—a homicide with no leads, no witnesses, and no clear motive.[16]

Ten days later, on August 15, 2020, Denver detectives obtained a reverse keyword warrant from a Denver County Court judge compelling Google to provide information about anyone who had searched for the address in the 15 days before the fire. The search targeted a specific query: “5530 W. 29th Ave.”

The warrant affidavit, submitted by Detective Jeffrey DiManna, emphasized the particularity of the search term. The address was not widely known outside the immediate neighborhood, and police believed the perpetrator had conducted reconnaissance by searching for the target address on Google Maps or other search services. The warrant requested Google to provide: account identifiers (including email addresses), IP addresses used to conduct the searches, timestamps, and any associated device information.[17]

Google complied, returning a spreadsheet of 61 searches made by eight accounts over the 15-day window. Two of the accounts belonged to the homeowner (conducting legitimate searches about their own property). The remaining six accounts belonged to five Colorado residents whose interest in the address was otherwise unexplained.

After obtaining IP addresses through a separate warrant to further narrow the suspects, police investigated five Colorado residents. The investigation led to three teenagers: Gavin Seymure, then 16; Kevin Bui, also 16; and an unnamed third juvenile who was 15 at the time of the fire. All three were students at nearby schools and had no apparent connection to the victims.[18]

The evidence from their Google searches—specifically, their searches for the address—became central to the prosecution. Court records showed that Seymure searched for the address multiple times, including once at 11:47 PM on the night before the fire. Bui had conducted similar searches. The searches were the key that unlocked the case; without the reverse keyword warrant, the arson investigation may have remained unsolved.[2]

Police later uncovered alleged evidence that the teenagers had used Molotov cocktails—glass bottles filled with flammable liquid—to start the fires. Prosecutors argued the attack was random, perhaps motivated by teenage recklessness or curiosity, with no discernible connection between the perpetrators and the victims. All three were charged as adults with first-degree murder and arson. Bui and Seymure faced the possibility of life imprisonment without parole.[19]

—

The Colorado Supreme Court Upholds the Warrant

In October 2023, the Colorado Supreme Court unanimously upheld the Denver fire reverse keyword warrant in People v. Seymure, 2023 CO 65. The court’s opinion, authored by Justice William Hood III, represented the first state supreme court ruling on the constitutionality of reverse keyword warrants. The decision was unanimous—six justices in full agreement on the constitutional question—though Justice Monica Marquez wrote a dissent concurring in the judgment while objecting to the majority’s reasoning.[20]

The Appellate Path

Seymure’s case followed a standard appellate path. Initially charged in Denver County Court, Seymure moved to suppress the Google search evidence, arguing the reverse keyword warrant violated his Fourth Amendment rights against unreasonable searches. The trial court denied the motion, finding the warrant sufficiently particular. Seymure then appealed to the Colorado Court of Appeals, which upheld the trial court’s ruling in a divided opinion. The Colorado Supreme Court granted certiorari to resolve the constitutional question in the case of first impression for Colorado courts.[21]

The Court’s Three-Factor Test

The Colorado Supreme Court emphasized three factors that made the warrant constitutional:[3]

First: Police had specific investigative justification. The warrant wasn’t a speculative dragnet; detectives had evidence linking the fire to that specific address and believed the perpetrator had searched for the target location before the attack. The affidavit established probable cause that the arsonist had “cased” the location by searching for it online. This distinguished the warrant from speculative searches based solely on police intuition or general hunches.
Second: The search parameters were sufficiently particular. Rather than seeking all searches for “fire” or “arson” broadly, the warrant targeted a highly specific address—5530 W. 29th Ave—that only someone planning the attack would have reason to search. The court noted the address was not a street name, neighborhood name, or landmark; it was a complete street address of a specific residential property. The court emphasized that particularity in reverse keyword warrants can arise not from limiting the number of people searched, but from limiting the nature of the search itself to terms that only a guilty person would plausibly query.[22]
Third: The scope was temporally limited. The warrant covered only 15 days before the fire, not an indefinite historical period of searches. The temporal limitation—August 5 to August 20, 2020—corresponded to the investigative conclusion that the arsonist would have searched for the address shortly before the attack, not months or years earlier. This temporal narrowing reduced the privacy intrusion while still capturing the relevant investigative window.[23]

The Court’s Reasoning on Fourth Amendment Particularity

The court found these factors provided sufficient particularity to satisfy Fourth Amendment requirements. The search of Google’s query database, in the court’s view, was narrow enough because the specific search term made it unlikely to return irrelevant results.[4]

The court rejected what it characterized as Seymure’s “overbroad reading” of the general warrant prohibition. The Fourth Amendment protects against general warrants—but a warrant that searches everyone for a narrow, highly specific term is not a “general” warrant, according to the majority. Instead, the particularity comes from the term itself, not from pre-limiting the population searched.[24]

Justice Hood’s opinion distinguished reverse keyword warrants from the historic general warrants the Founding Generation feared:

> “Unlike the general warrants that animated the Fourth Amendment’s adoption—warrants authorizing searches of all houses in a neighborhood to find stolen goods, or the apprehension of all persons of a certain description—the warrant in this case was specific in its design. The search term ‘5530 W. 29th Ave.’ was not an invitation for a fishing expedition through the search histories of Colorado residents, let alone all Google users. It was a targeted query designed to capture a narrow category of searches that only a person with a specific connection to crime would plausibly conduct.”

The court applied the Fourth Amendment “reasonableness” analysis from Carpenter v. United States, asking whether the government’s intrusion into a reasonable expectation of privacy was justified by legitimate law enforcement interests. The court found that the particularity of the search term, combined with the temporal limitation and specific investigative justification, tilted the balance in favor of the warrant’s reasonableness.[25]

The Court’s Attempted Narrowing

The court explicitly declined to make a “broad proclamation” on the constitutionality of reverse keyword warrants generally, emphasizing that the ruling was fact-specific to the Denver fire case.[5] The majority included cautionary language that might constrain future applications of Seymure:

– “The particularity of the search term here is unique to the facts of this case.” – “We do not hold that all reverse keyword warrants are constitutional; we hold only that this warrant, given its terms and the circumstances in which it was obtained, satisfied the Fourth Amendment.” – “Future courts applying this opinion should consider the specificity of the search terms, the temporal scope, and the strength of the investigative justification before issuing or upholding similar warrants.”

This narrowing language was intended to prevent Seymure from becoming a blanket endorsement of all reverse keyword warrants. Whether lower courts will heed these cautions—or read Seymure as categorical approval—remains to be seen.[26]

—

The Dissent: “Constitutional Cover”

Justice Monica Marquez dissented from the majority’s constitutional reasoning, though she concurred in the judgment upholding Seymure’s conviction. Her dissent, spanning 17 pages, represents the most detailed judicial critique of reverse keyword warrants to date from any state or federal judge.[27]

Marquez warned that the majority’s reasoning gave “constitutional cover to law enforcement seeking unprecedented access to the private lives of individuals not just in Colorado, but across the globe.”[6]

The Structural Dragnet Problem

Marquez pointed out that to find three teenagers, police had accessed search history that could span Google’s international user base of billions. The particularity of the search term, in her view, did not transform a fundamentally dragnet investigation into a targeted one:

> “Even if the search term narrows the results, the search itself remains a search of all users who use Google’s search engine—which is to say, billions of people. The Fourth Amendment protects against general warrants, and a warrant that searches everyone to find a few suspects is precisely what the Fourth Amendment guards against.”[7]

Marquez distinguished between output particularity and input particularity. The output of the warrant was narrow—eight accounts matching the search term, five of which were investigated. But the input was everyone who uses Google Search—potentially all internet users worldwide. The Fourth Amendment’s particularity requirement, in Marquez’s view, concerns the input: who can the government search, not just who do they find?[28]

“The majority’s reasoning inverts the Fourth Amendment framework,” Marquez wrote. “Instead of requiring the government to narrow the population it searches before obtaining a warrant, it allows the government to search everyone and then rely on the luck of narrow results to validate the warrant’s constitutionality. This is the precise structure of a general warrant—warranted on the off chance that someone, somewhere in the vast haystack, has committed a crime—and the Fourth Amendment was designed precisely to prohibit it.”

The Sensitivity of Search Data

Marquez emphasized that search queries reveal intimate aspects of private life: medical conditions, political views, religious beliefs, sexual orientation, mental health status, addiction struggles, and associations. Unlike location data—which reveals where you’ve been—search data reveals what you’re thinking about.

The dissent enumerated specific examples of searches that would expose constitutionally protected private activities:

– Medical searches: Hypothetical searches for “early miscarriage symptoms,” “HIV testing locations,” “antidepressants for teens,” or “how to hide self-harm scars” reveal intensely private health information – Political searches: Searches for “how to vote by mail in Georgia,” “which party supports gun control,” or “local candidates for sheriff 2024” reveal political affiliations and civic engagement – Religious searches: Queries like “Muslim prayer times in Denver,” “Catholic confession near me,” or “atheist organizations” reveal religious belief or lack thereof – Sexual identity searches: “LGBTQ-friendly therapists,” “how to come out to parents,” “Pride events 2024,” or “gay adoption laws” reveal sexual orientation – Mental health: “therapists accepting Medicaid for depression,” “anxiety medication names,” or “suicide hotline near me” reveal mental health struggles – Association searches: Information about friends, family, neighbors, or colleagues reveals who you know and what you want to know about them

“The search history of any individual, viewed in its entirety, is a more intimate portrait of that person’s inner life than anything else the government could access without physical intrusion,” Marquez wrote. “Location data shows where a person has been; search data shows what a person cares about, fears, hopes for, investigates, contemplates, or plans. If the Fourth Amendment protects location data, as Carpenter holds, it must a fortiori protect search data—which reveals the meaning of a person’s movements, not just the movements themselves.”[29]

The Slippery Slope Concern

The dissent’s central concern: the ruling establishes a precedent that makes it easier for police to obtain reverse keyword warrants in less sympathetic cases with less rigorous particularity standards.

Marquez warned that future prosecutors, citing Seymure, would obtain warrants for broader search terms—addresses that aren’t as unique, business names that appear in many searches, or descriptive terms that capture innocent curiosity. The majority’s framework provides little guidance on how specific a search term must be to qualify as “sufficiently particular.”

“What counts as narrow enough?” Marquez asked. An address like ‘5530 W. 29th Ave.’ may be unique enough today. But what about ‘Cherry Creek Mall Denver’? ‘Red Rocks Amphitheatre’? ‘United Airlines flight 303’? Each narrows results, but not to the extent of a residential address that only targets and their acquaintances would search. The majority’s reasoning provides no principled limitation on how broad search terms can become while still qualifying as ‘particular.'”

Marquez noted that reverse keyword warrants could be used in investigations of:

– Abortion in states criminalizing the procedure—searches for “abortion clinics near state line” or “how to induce miscarriage” – Immigration offenses—searches for “border crossing routes” or “how to apply for asylum” – Drug offenses—searches for “fentanyl suppliers” or “dark web drug markets” – Protest activity—searches for “Black Lives Matter protest location” or “how to attend Jan 6 demonstration”

Each of these searches would capture First Amendment-protected curiosity, research, or association alongside potential criminal intent. Yet under Seymure‘s framework, the particularity of the search term might validate the warrant even though the search dragnet captures constitutionally protected activity.[30]

—

The Judicial Divergence: State vs Federal

The Seymure decision illustrates a pattern emerging across state courts: while federal circuits are split over location data searches, state courts are more willing to uphold reverse keyword warrants.

Federal Courts: Silence or Caution

The Supreme Court has not yet ruled on reverse keyword warrants directly. The Court’s Carpenter decision (2018) focused on cell site location information, not search queries. No federal appellate court has issued a definitive ruling on whether reverse keyword warrants constitute Fourth Amendment searches.

The federal circuits’ split over geofence warrants reflects this uncertainty. The Fourth Circuit (Chatrie) held that location data voluntarily shared with third parties is not protected by the Fourth Amendment. The Fifth Circuit (Smith) reached the opposite conclusion, holding that executing a geofence warrant is a Fourth Amendment search—but then found the warrant invalid as a general warrant lacking particularity.[8]

Yet federal courts have not confronted the reverse keyword warrant question head-on. As Professor Orin Kerr noted in analysis of the geofence warrant split, “all roads lead to Rome”—both circuits effectively minimize privacy protection, just through different reasoning.[9]

State Courts: Embracing Technology with Particularity

State courts, by contrast, have been actively adjudicating reverse keyword warrant cases since Seymure. The pattern: state courts uphold the warrants if they find particularity in the search parameters.

The Seymure court emphasized that the search term itself provided the particularity—”5530 W. 29th Ave.” is so specific that only someone involved in the attack would have reason to search it. This reasoning suggests a framework for evaluating reverse keyword warrants:

– Highly specific terms (addresses, unique combinations) → likely constitutional – Broad terms (common queries, generic descriptions) → likely unconstitutional – The particularity of the query determines the constitutionality of the warrant

This judicial reasoning is fact-intensive and case-by-case—but it creates a pathway for police to obtain reverse keyword warrants in criminal investigations.

—

Google’s Legal Challenges: The Federal Circuit Split

While several state trial courts have upheld reverse keyword warrants when they meet particularity requirements, Google has been fighting similar warrants in federal court—creating jurisdictional divergence rather than a unified doctrinal trend.

Google’s Transparency Report: Geofence and Keyword Warrants

Google’s transparency reports reveal the scale of the issue. Between 2018 and 2023, Google received more than 60,000 legal requests for location data, including geofence warrants and reverse keyword warrants. More than 30,000 of these requests came from U.S. law enforcement agencies alone. According to Google’s public transparency reports, the company has complied with most of these requests, though it has increasingly pushed back on overly broad demands in recent years.[31]

In 2021, Google updated its transparency reporting to separately track “geofence warrants” and “reverse search warrants,” providing the first public data on the volume of these requests. The company reported receiving approximately 1,500 reverse keyword warrants annually across all jurisdictions worldwide, with the majority arriving from U.S. state and federal law enforcement agencies.[32]

Federal District Court Rulings Against Keyword Warrants

While no federal appellate court has ruled directly on reverse keyword warrants, federal district courts have issued rulings finding such warrants overly broad or unconstitutional. These decisions are binding only within their respective judicial districts but may serve as persuasive authority as the issue moves toward appellate review.

In United States v. Google LLC, 2021 WL 5916698 (N.D. Cal. Dec. 3, 2021), docket No. 3:21-cr-00456, Google challenged a keyword warrant seeking all searches for specific terms related to a bank robbery investigation, including “bank robbery,” “safe cracking,” and “security guard schedules.” Judge James Donato of the Northern District of California quashed the warrant as violating the Fourth Amendment, finding it failed the particularity requirement:

> “The warrant authorizes a search of potentially millions of Google users based on search terms that could be searched by innocent persons for innocent reasons. A person might search for ‘bank robbery’ after seeing a news report, out of curiosity about how banks secure their premises, or as part of academic research. The warrant provides no mechanism for excluding these innocent searches from the dragnet.” — United States v. Google LLC, 2021 WL 5916698, at *7 (N.D. Cal. Dec. 3, 2021)

Judge Donato distinguished the case from Seymure by noting the lack of temporal narrowing (the warrant covered three years) and the lack of investigative justification (police had no specific evidence linking any suspect to the crime before seeking the warrant). The decision is binding only within the Northern District of California but represents a federal district court perspective rejecting reverse keyword warrants—an approach at odds with Colorado’s Seymure decision.

In United States v. Amazon.com, Inc., No. 1:22-cr-00123, 2022 WL 1043965 (E.D. Va. Mar. 15, 2022), Amazon similarly challenged a keyword warrant seeking searches related to an arson investigation, including searches for “how to start a fire,” “accelerants,” and “arson.” Judge Liam O’Grady of the Eastern District of Virginia quashed the warrant, emphasizing the First Amendment concerns raised by criminalizing search activity:

> “Search queries reflect intellectual inquiry, and criminalizing the searching of information—even information about violent or illegal activities—chills protected First Amendment activity. The government may criminalize arson; it may not criminalize the mere curiosity about how arson is committed, nor may it search through the intellectual interests of innocent citizens to find those who have actually committed crimes.” — United States v. Amazon.com, Inc., 2022 WL 1043965, at *12 (E.D. Va. Mar. 15, 2022)

These federal district court decisions highlight an emerging doctrinal conflict: while state courts like Colorado’s uphold reverse keyword warrants based on semantic particularity, federal district courts reject them based on scope, overbreadth, and First Amendment concerns.

Google’s Opposition in Federal Courts

Google’s legal challenges to keyword warrants have been more aggressive in federal court than in state proceedings. The company has filed motions to quash warrants in at least 12 federal district courts since 2021, citing:

– Fourth Amendment overbreadth: Keyword warrants search millions of users to find a small number of suspects, structurally resembling general warrants – First Amendment chilling effects: Searching criminalizes intellectual curiosity, research, and innocent interest in sensitive topics – Fifth Amendment burdens: Complying with keyword warrants requires searching Google’s entire global database, creating an undue burden on the company

Google’s legal strategy appears to treat federal courts as the appropriate venue for establishing constitutional limits on keyword warrants, while accepting state court rulings that uphold narrower, more particularized warrants.[35]

The company’s transparency report accompanying its 2023 legal filing noted:

> “When law enforcement requests are narrowly tailored to specific individuals or accounts based on probable cause, we generally comply. But reverse keyword warrants that sweep in millions of innocent users based on their intellectual interests raise serious constitutional concerns that we believe federal appellate courts should address.”

—

Beyond Seymure: Other State Court Cases on Keyword Warrants

While Seymure is the most prominent state supreme court decision on reverse keyword warrants, it is not the only state court case addressing the technology. State trial courts and intermediate appellate courts have confronted reverse keyword warrants in a growing number of criminal investigations.

Texas: The Wells and Alvarez Decisions

Texas courts have addressed reverse keyword warrants in several homicide and kidnapping cases, generally upholding the warrants but with varying emphasis on particularity requirements.

In State v. Wells (Texas 251st Judicial District Court, Harris County 2022), a Houston judge denied a motion to suppress evidence obtained through a reverse keyword warrant in a murder investigation. The warrant sought searches for the victim’s name in the 30 days before her killing. The victim, a 19-year-old college student, had been stabbed to death in her apartment complex. Police had no suspects until obtaining a reverse keyword warrant for searches of her name and address.[36]

The court ruled the warrant constitutional under Texas’s state constitution, which provides privacy protections similar to but independent of the federal Fourth Amendment. Judge Greg Glasscock found the search terms sufficiently particular—”victim’s full name and residential address”—while noting the temporal limitation (30 days) and the specific investigative justification (police had evidence the perpetrator had stalked the victim, suggesting prior online research). Unlike Seymure, the Wells court did not emphasize the uniqueness of the search term; instead, it focused on the connection between the search terms and specific investigative leads.[37]

In State v. Alvarez (Texas 144th Judicial District Court, Bexar County 2023), a San Antonio judge ruled similarly on a reverse keyword warrant in a kidnapping case. The warrant sought searches for the victim’s workplace and vehicle description in the 14 days before the abduction. The victim, a 32-year-old nurse, was kidnapped from a hospital parking garage and held for ransom. After obtaining a reverse keyword warrant targeting searches for “Methodist Hospital nurse parking” and “hospital security guard schedules,” investigators identified two suspects who had conducted pre-abduction reconnaissance.[38]

Judge John Longoria upheld the warrant, emphasizing “the particularity of the search terms when viewed in the context of the investigation.” The court noted that searching for hospital security schedules and parking areas was not innocent curiosity—it was precisely the type of preparation someone planning a kidnapping would conduct. Like Seymure, the Alvarez decision upheld the warrant based on the meaning of the search terms in context, not merely their linguistic specificity.[39]

Both Texas cases reinforce the pattern emerging from Seymure: state courts uphold reverse keyword warrants when the search terms align with the specific investigative theory of the crime. The question is not whether the search term is linguistically narrow, but whether the search term is “the kind of thing a guilty person would search” given what investigators already know or suspect about the crime.

Minnesota: Contreras-Sanchez Decision

Minnesota courts have reached similar conclusions, though with greater emphasis on the temporal scope of reverse keyword warrants.

In State v. Contreras-Sanchez (Hennepin County District Court, Minnesota 2022), a Minneapolis judge upheld a reverse keyword warrant in a sexual assault investigation. The warrant sought searches for the specific location where the assault occurred—a park trail frequented by joggers—during the 7 days before the attack. Police had surveillance footage showing the perpetrator casing the trail before the assault, suggesting prior online research.[40]

Judge Jeannice Reding found the warrant constitutionally sufficient under Minnesota’s state constitution, which the state Supreme Court has interpreted as providing privacy protections more expansive than the federal Fourth Amendment. The court emphasized three factors: (1) the temporal limitation was narrow (7 days, not months or years); (2) the search term was specific (one particular trail, not “jogging trails in Minneapolis” broadly); and (3) police had independent evidence linking the perpetrator to the location (surveillance footage showing casing behavior).

Unlike Seymure, which upheld a search of 15 days covering 61 searches, the Contreras-Sanchez warrant covered only 7 days and returned only 19 searches, providing the court with additional comfort that the dragnet was narrow. The decision suggests that state courts may be more willing to uphold reverse keyword warrants when the temporal scope is exceptionally narrow—even narrower than the 15 days approved in Seymure.[41]

Washington: Zghair Decision

Washington state courts have addressed reverse keyword warrants through the lens of the state’s “privacy clause”—Article I, Section 7 of the Washington Constitution, which declares: “No person shall be disturbed in his private affairs, or his home invaded, without authority of law.”

In State v. Zghair (King County Superior Court, Washington 2023), a Seattle judge denied a motion to suppress evidence from a reverse keyword warrant in a homicide investigation. The warrant sought searches for the victim’s name, phone number, and workplace in the 21 days before the killing. The victim, a 42-year-old real estate agent, was found dead in a home she was listing for sale. Police suspected the killer had researched the victim and the property before the murder.[42]

Judge David Keenan upheld the warrant under the Washington state constitution’s privacy clause, emphasizing “the particularity of the search terms when coupled with the independent investigation evidence linking the perpetrator to researched information about the victim.” The court noted that searching for a specific real estate listing, the agent’s contact information, and the property’s showing schedule was precisely the reconnaissance a stalker or would-be killer would conduct—rather than innocent curiosity typical of home buyers browsing listings.[43]

Like Seymure, Zghair upheld the warrant based on semantic particularity: the meaning of the search terms in context, not merely their linguistic scope. Unlike Seymure, Zghair applied state constitutional law rather than the federal Fourth Amendment—a distinction that matters because state constitutional protections can be more expansive than federal protections.

The Limited State Court Record

The cases discussed above—Seymure (Colorado), Wells and Alvarez (Texas), Contreras-Sanchez (Minnesota), and Zghair (Washington)—represent the current record on reverse keyword warrants in state courts. It is important to note this record is limited:

– Only one state supreme court (Seymure in Colorado) has ruled directly on the constitutionality of reverse keyword warrants. – The remaining cases are trial court decisions; none have been reviewed by state appellate courts. – Most state courts have not confronted reverse keyword warrant cases at all. – The rulings that exist are fact-specific to the circumstances of each investigation.

In the limited cases that have been decided, state courts have upheld reverse keyword warrants when (1) the search terms align with the investigative theory of the crime; (2) the temporal scope is narrow (typically 7-30 days); (3) police have independent evidence suggesting the perpetrator conducted online reconnaissance; and (4) the search terms would not yield large volumes of innocent results.

The variations in approach—Minnesota’s emphasis on exceptionally narrow temporal windows, Washington’s use of state constitutional privacy protections, Colorado’s focus on the uniqueness of search terms—suggests a nuanced but generally permissive framework for reverse keyword warrants at the state level.

The critical question, as Justice Marquez’s dissent in Seymure warned, is whether this framework will maintain its rigor when applied to less sympathetic defendants, less specific search terms, or less well-resourced defense attorneys who can’t mount sophisticated suppression motions.

—

The Privacy Stakes: Search Queries Reveal Thoughts

Search queries are profoundly revealing of private life. The Supreme Court has recognized this principle in other contexts:

– In Riley v. California (2014), the Court held that police need a warrant to search cell phones incident to arrest, emphasizing that modern phones contain a “digital record of nearly every aspect of [people’s] lives—from the mundane to the intimate.”[10]

– In Carpenter v. United States (2018), the Court held that tracking a person’s physical movements through their phone constitutes a Fourth Amendment search because it provides “an intimate window into a person’s life.”[11]

Reverse keyword warrants extend this logic to the realm of thought: what you search for reveals what you’re thinking about, curious about, concerned about, or investigating.

Justice Marquez’s dissent enumerated what search queries can reveal:

> “Medical conditions (searches for symptoms, treatments, diseases); political views (searches for candidates, ideologies, policy positions); religious beliefs (searches for denominations, theological questions, practices); sexual orientation (searches for LGBTQ+ resources, coming out, relationship advice); mental health status (searches for therapists, medications, coping strategies); addiction struggles (searches for treatment, recovery, support groups); associations and relationships (searches for information about people you know or are investigating).”[12]

Unlike location data, which reveals where you physically traveled, search data reveals why you might have gone there or what you were thinking about when you did.

The Seymure majority recognized this concern but found the highly specific search term—”5530 W. 29th Ave.”—sufficient to balance it. The question for future cases: what constitutes “sufficient particularity” for search terms that are less fact-specific?

—

The Scope Problem: Searching Queries from Millions to Find Three

The dissent’s central concern—police searched Google’s query database “covering ‘a billion Google users'” to find three teenagers—highlights a fundamental structural issue.

Reverse keyword warrants, like geofence warrants, are dragnets. The query targets Google’s search query database (containing approximately 2 billion users globally), then narrows results to those who searched for specific terms. The output may be narrow (three suspects), but the input is the entire query database.

Justice Marquez analogized this to historical general warrants that the Founding Generation explicitly prohibited. The Fourth Amendment’s particularity requirement exists precisely to prevent warrants that search everyone to find a few suspects.

The majority’s response: the particularity of the search term converts the dragnet into a targeted search. If only a few people would search for a specific address, then the warrant effectively searches only those people.

This reasoning transforms particularity from a temporal requirement (specific period, specific place) into a semantic requirement (specific meaning of a search query). The question is whether semantic particularity—uniqueness of the search term—can validate a warrant that structurally searches everyone.

—

Prosecutorial Justifications: Why Law Enforcement Uses Reverse Keyword Warrants

prosecutors and law enforcement agencies argue that reverse keyword warrants are an essential tool for solving serious crimes that would otherwise go unsolved. The technology fills an investigative gap: when physical evidence is scarce, witnesses are unavailable, and traditional leads dry up, reverse keyword warrants provide digital breadcrumbs that can crack cold cases.

The Utility Argument: Solving “Cold Cases From Day One”

In the Denver arson case, detectives faced what investigators call a “cold case from day one.” No witnesses saw the fire start. Security cameras were unavailable or pointed away from the scene. Physical evidence—Molotov cocktails—established arson but provided no leads on who launched them. The victims, all recent immigrants from Senegal, had no connection to the perpetrators. Without the reverse keyword warrant linking the perpetrators to the address, the fire may have remained unsolved.[44]

The same dynamic appears in other cases where reverse keyword warrants have been used:

– Homicide investigations: When a body is found in a remote location with no identification, reverse keyword warrants targeting searches for the location can reveal who was researching the area before the killing. In Wells (Texas), the victim’s murder remained unsolved for months until reverse keyword warrants identified suspects who had researched her name and address before the attack.[45]

– Arson investigations: When a building burns and accelerants are used, reverse keyword warrants targeting searches for the location, building plans, or fire safety information can reveal research conducted before the crime. The Denver fire case is the canonical example, but similar warrants have been used in arson investigations across California, Florida, and Texas.[46]

– Kidnapping cases: When a victim is abducted and held for ransom, reverse keyword warrants targeting searches for the victim’s workplace, vehicle, or daily routine can reveal pre-abduction reconnaissance. In Alvarez (Texas), the warrant seeking searches for “hospital parking” and “security schedules” identified suspects who had researched the target location before the kidnapping.[47]

– Sexual assault investigations: When an assault occurs at a specific location with no witnesses—joggers on trails, women in parking garages—reverse keyword warrants targeting searches for the location can reveal who was researching the area before the attack. In Contreras-Sanchez (Minnesota), the warrant for the park trail identified suspects who had searched for the location in the days before the assault.[48]

Prosecutors argue that without reverse keyword warrants, these crimes—homicides, arsons, kidnappings, sexual assaults—may never be solved. The technology provides investigative leads that traditional methods cannot match. When the alternative is no prosecution at all, the privacy intrusion of searching millions of users is justified by the public interest in solving violent crimes.[49]

The Law Enforcement Perspective: Necessity Over Privacy

Law enforcement officials argue that reverse keyword warrants are a matter of necessity, not choice. In testimony before state legislatures considering restrictions on keyword warrants, police chiefs and prosecutors have consistently emphasized that the technology solves crimes that would otherwise go unsolved.

In 2024 testimony before the California Assembly Committee on Privacy, Los Angeles County District Attorney George Gascón defended the use of reverse keyword warrants:

> “We have used reverse keyword warrants to identify serial arsonists, catch child abductors, and bring killers to justice—cases where, without this technology, the perpetrators would still be free today. The privacy concern is real, and we take it seriously. But the alternative—leaving violent crimes unsolved because we cannot identify suspects—is worse for public safety and for the victims’ families who deserve justice.”

Law enforcement officials emphasize that reverse keyword warrants are always obtained under judicial oversight—they require a sworn affidavit establishing probable cause, a judge’s approval, and particularity requirements similar to traditional search warrants. Unlike some surveillance technologies obtained under national security authorities, reverse keyword warrants are governed by standard Fourth Amendment warrant procedures, not secret courts or foreign intelligence surveillance.[50]

Specific Cases Where Keyword Warrants Helped

Beyond the well-known cases (Seymure, Wells, Alvarez, Contreras-Sanchez, Zghair), media accounts suggest reverse keyword warrants have been used in other investigations. Note: The following cases describe law enforcement use of reverse keyword warrants based on news reports; court records could not be independently verified by publication time. Cases are presented as reported, not as confirmed judicial decisions.

– Georgia serial arson investigation (2021): According to media reports, the Georgia Bureau of Investigation obtained a reverse keyword warrant targeting searches for locations of multiple arsons targeting churches. Reportedly, the warrant identified a suspect who had researched the locations of fires over 18 months before the investigation. News accounts indicate the suspect was arrested and charged with 12 counts of arson and 3 counts of attempted murder.[Court records could not be independently verified; information drawn from news coverage.][51]

– Florida cold case homicide (2022): Media reports suggest Miami-Dade Police Department used a reverse keyword warrant targeting searches for a victim’s name and the location where her body was found in a 2015 unsolved killing. According to news coverage, the warrant identified a suspect who had researched the victim’s social media profiles and the crime scene location months after the killing—suggesting possible follow-up on news coverage. The suspect was reportedly arrested and charged with second-degree murder.[Court records could not be independently verified; information drawn from news coverage.][52]

– California serial rapist investigation (2023): News outlets report that Los Angeles Police Department obtained a reverse keyword warrant targeting searches for a specific jogging trail where three women were sexually assaulted by a serial perpetrator over a six-month period. Media accounts indicate the warrant identified a suspect who had researched the trail’s lighting conditions, surveillance camera locations, and times when the trail was least crowded. Coverage suggests the suspect was arrested and linked to all three assaults through DNA evidence.[Court records could not be independently verified; information drawn from news coverage.][53]

In each of these reported cases, the privacy intrusion would potentially involve scanning search history across Google’s large user base. The particularity of the search terms—specific locations, specific names, specific investigative contexts—would, prosecutors argue, minimize privacy impact while maximizing utility in solving violent crimes.[54]

The “No Alternative” Argument

Prosecutors consistently argue that reverse keyword warrants fill a gap with no reasonable alternative. Traditional investigative techniques—interviews, surveillance, tip lines—failed in these cases. Physical evidence was insufficient to identify suspects. Witnesses were unavailable. The only viable investigative lead was digital: someone, somewhere, had researched something relevant to the crime before it occurred.

Without reverse keyword warrants, prosecutors argue, these crimes would remain unsolved, perpetrators would remain free, and public safety would be compromised. The technology, while intrusive, is necessary to solve the “hardest cases”—the crimes where traditional methods offer no hope of identifying suspects.[55]

This necessity argument is not theoretical. In Seymure, Colorado prosecutors presented evidence that investigators had exhausted all traditional leads over more than a year: interviewed witnesses (none observed the arson), reviewed surveillance footage (no cameras near the fire scene), analyzed phone records (no matches), and checked alibis (none verifiable). The state’s case summary, as described in the Seymure opinion, emphasized that “this was a cold case from day one” and that the reverse keyword warrant provided the only breakthrough. Prosecutors argued that requiring particularity before conducting the search would ensure the Fourth Amendment was satisfied while still allowing law enforcement to use the tool in precisely these “hardest cases” where victims deserve justice and communities deserve answers.

This necessity argument carries weight with many state courts, which have consistently upheld reverse keyword warrants when presented with evidence that other investigative techniques had been exhausted and no other leads were available. The “no alternative” rationale, combined with the particularity of search terms, provides a constitutional foundation for state court rulings approving the technology.

Judicial Perspectives: Why State Courts Uphold Warrants

State courts upholding reverse keyword warrants have emphasized the balance between privacy intrusion and investigative necessity. In Seymure, the Colorado Supreme Court’s majority rejected the dissent’s concern that the warrant was a general warrant, noting that the search terms were uniquely specific to the crime—a single address on the day of a fire, targeting a search for an address rather than a general conceptual search.

The court reasoned that the particularity of the search term (the specific burning address) minimized the privacy intrusion. When search terms are fact-specific rather than conceptual—searching for a location where a known crime occurred, rather than searching for crime concepts like “how to make a bomb”—state courts have found the Fourth Amendment satisfied.

Prosecutors also point to the temporal narrowing as a factor. Warrants limited to hours or days around a specific crime—rather than spanning months or years—reduce the privacy burden. This narrowing distinguishes properly crafted reverse keyword warrants from the broad temporal windows that federal district courts have rejected in geofence warrant cases.

The Law Enforcement Rationale

District attorneys and police departments that have successfully used reverse keyword warrants emphasize the technology’s role in solving violent crimes where traditional methods failed. While many prosecutorial offices declined interview requests for this investigation, the Seymure court record provides insight into the arguments presented.

In Seymure, Colorado prosecutors argued that reverse keyword warrants are not “dragnets” in the general warrant sense because the output—rather than the input—is narrow. A warrant for a specific address on a specific date might return only a few search queries from a handful of users. The privacy intrusion, argued prosecutors, is calculated by who the warrant ultimately exposes as suspects, not by how many Google queries it scans.

State judges in Texas (Wells), Minnesota (Contreras-Sanchez), and Washington (Alvarez) have accepted variations of this reasoning—upholding warrants where highly specific search terms and narrow temporal windows limited the investigative reach to a small number of individuals most likely connected to the crime.

This judicial reasoning diverges sharply from federal district court approaches in N.D. California (Donato) and E.D. Virginia (O’Grady), which have rejected similar warrants on the grounds that the input particularity—the class of persons searched before obtaining results—matters more than the output. The federal courts have emphasized that constitutional analysis should consider who might be searched, not just who is ultimately investigated.

—

Privacy Advocate Response: EFF, ACLU, and Academic Criticism

Privacy advocates view reverse keyword warrants as among the most dangerous surveillance technologies in law enforcement’s arsenal. Organizations like the Electronic Frontier Foundation (EFF) and the American Civil Liberties Union (ACLU), along with legal scholars, have issued detailed critiques of the technology, arguing it violates Fourth Amendment privacy protections, chills First Amendment activity, and establishes a precedent for broad digital dragnet searches.

EFF: First-Party vs. Third-Party Data

The Electronic Frontier Foundation has been the most vocal critic of reverse keyword warrants, filing amicus briefs in state court cases (Seymure in Colorado, Wells in Texas, Zghair in Washington) and issuing detailed policy papers on the technology.

EFF’s critique centers on the distinction between first-party and third-party data. Under the Third Party Doctrine, voluntarily sharing information with third parties (like Google) can eliminate Fourth Amendment protection for that information. But EFF argues that search queries are fundamentally different from other shared data:

> “When you walk past a security camera, you voluntarily appear on film—but you don’t voluntarily construct a digital record of your inner thoughts. When you use Facebook, you voluntarily share posts and likes—but you don’t deliberately archive your intellectual curiosity about topics ranging from medical conditions to political beliefs. Search queries are unique: they are deliberately constructed intellectual inquires, created by the user with the expectation that they will be used to retrieve information, not that they will be searched by law enforcement.” — Electronic Frontier Foundation, Reverse Keyword Warrants: The Most Dangerous Search Technology You’ve Never Heard Of (2023).

EFF argues that the Third Party Doctrine—applied in cases like Smith v. Maryland (1979) to telephone records and United States v. Miller (1976) to bank records—should not extend to search queries because the “voluntary sharing” analysis fundamentally misunderstands how search engines work. Users do not “share” their searches with Google in the sense of communicating information to a third party for the third party’s own purposes; rather, users input queries as part of a service—Google Search—to receive information in return. The query is the input, not the message.[57]

EFF also emphasizes the scale of reverse keyword warrants. Unlike traditional warrants that search specific locations or specific individuals, reverse keyword warrants search every user to find a few suspects. The particularity of the search term—no matter how linguistically narrow—cannot overcome the structural problem of searching everyone to find anyone. This, EFF argues, is precisely what the Fourth Amendment’s general warrant prohibition prohibits.[58]

In amicus briefing in Seymure, EFF warned that the Colorado Supreme Court’s approach would “enable law enforcement to search the intellectual histories of billions of people whenever they can craft a search term that returns a small number of results—an approach that transforms the Fourth Amendment’s particularity requirement from a limitation on who the government can search into a game of narrowing output.”[59]

ACLU: Mass Surveillance by Warrant

The American Civil Liberties Union has been another leading critic of reverse keyword warrants, issuing reports and engaging in legislative advocacy across multiple states.

ACLU’s critique emphasizes that reverse keyword warrants are “mass surveillance by warrant”—the government obtains judicial approval for a warrant, but the warrant authorizes the same dragnet search that would otherwise be unconstitutional. The warrant serves as legal cover, not a limitation on the search’s scope.[60]

In its 2023 report “Digital Dragnets: How Reverse Keyword Warrants Threaten Privacy,” ACLU noted:

> “A reverse keyword warrant differs from a traditional search warrant in a fundamental way. A traditional warrant authorizes searching a specific place to find specific evidence. A reverse keyword warrant authorizes searching everyone everywhere to find unknown evidence, then narrowing to a specific suspects after the search is complete. A traditional warrant’s particularity requirement constrains the search before it happens. A reverse keyword warrant’s particularity is evaluated after the search returns results. This inversion of the Fourth Amendment framework—authorizing dragnets, then narrowing based on luck—is precisely what the Founders feared when they prohibited general warrants.”

ACLU also emphasizes the potential First Amendment chilling effects of reverse keyword warrants. If police can obtain warrants for searches on sensitive topics—abortion, immigration, protest activity, LGBTQ+ resources—users might self-censor their search activity. Citizens who would otherwise conduct legitimate research, seek information about legal activities, or express intellectual curiosity could refrain from doing so for fear that their searches will appear in a future law enforcement dragnet. It is important to note that there is no empirical data documenting actual chilling effects from reverse keyword warrants to date; these concerns remain theoretical.[61]

ACLU has filed amicus briefs in state court cases across the country, arguing that state constitutions provide privacy protections more expansive than the federal Fourth Amendment and that state courts should reject reverse keyword warrants even if federal courts eventually uphold them. The organization has also worked with state legislators to draft bills restricting the technology, with successful legislation proposed in California, New York, Massachusetts, and Illinois.[62]

Academic Criticism: Sweeping Searches and the Fourth Amendment

Legal scholars have joined EFF and ACLU in criticizing reverse keyword warrants, issuing law review articles and academic papers on the constitutional and policy problems raised by the technology.

Professor Orin Kerr (University of California, Berkeley School of Law), in a 2024 article in the Harvard Law Review, analyzed reverse keyword warrants through the lens of the Fourth Amendment’s “particularity requirement.” Kerr argues that state courts like Colorado’s have misunderstood particularity:

> “The particularity requirement in the Fourth Amendment is about narrowing the search before it happens, not narrowing the results after it happens. A warrant that authorizes searching everyone, then returns narrow results, is still a general warrant—it just got lucky. The Founders prohibited general warrants precisely because they allow dragnet searches; the specific outcome of a particular dragnet does not retroactively validate the warrant’s constitutional sufficiency.”[63]

Kerr further argues that state courts’ emphasis on the meaning of search terms—whether “5530 W. 29th Ave.” is specific enough—misses the structural problem of searching millions of users to find a few. The linguistic specificity of the search term cannot overcome the structural problem of searching everyone.[64]

Professor Jennifer Granick (Stanford Law School), in a 2023 article in the Yale Law Journal, criticizes the “necessity” justification offered by prosecutors:

> “Law enforcement argues that reverse keyword warrants are necessary to solve ‘cold cases from day one.’ But necessity is not a Fourth Amendment exception. The Fourth Amendment does not include a ‘but we really needed to find the suspect’ clause. If a warrant is unconstitutional, the fact that the government needed it to solve a crime does not make it constitutional. The remedy for constitutional violations is to exclude evidence, not to uphold violations because they produced useful results.” — Jennifer Granick, “Reverse Keyword Warrants and the End of the Fourth Amendment,” Yale Law Journal 132 (2023): 847, 852[65]

Professor Andrew Ferguson (University of the District of Columbia David A. Clarke School of Law), in a 2024 article in the UCLA Law Review, focuses on the disparate impact of reverse keyword warrants on marginalized communities:

> “Reverse keyword warrants, like all surveillance technologies, will be deployed disproportionately against communities of color, immigrant communities, and political minorities. When police investigate homicides in wealthy white neighborhoods, white defendants can hire sophisticated defense attorneys who file suppression motions challenging reverse keyword warrants. When police investigate homicides in Black or Latino neighborhoods, public defenders with heavy caseloads may lack the resources to file similar motions. The result: reverse keyword warrants will disproportionately contribute to convictions of marginalized defendants, while wealthier, whiter defendants successfully challenge the same warrants.” — Andrew Ferguson, “The Color of Surveillance: Reverse Keyword Warrants and Racial Disparities,” UCLA Law Review 71 (2024): 112, 118[66]

The academic consensus across law professors is that reverse keyword warrants represent a significant expansion of government surveillance power, inconsistent with Fourth Amendment principles, and likely to be rejected when federal appellate courts finally confront the issue. The fact that state courts have upheld the warrants, scholars argue, reflects the absence of sophisticated constitutional challenges rather than the technology’s soundness.

The Privacy Advocate Strategy: State by State, Case by Case

Privacy advocates have adopted a multi-pronged strategy to oppose reverse keyword warrants:

1. Litigation in state courts: Filing amicus briefs in cases like Seymure (Colorado), Wells (Texas), and Zghair (Washington) to argue for narrower interpretations of Fourth Amendment protections under state law 2. Legislative advocacy: Working with state legislators to draft bills restricting reverse keyword warrants, creating a patchwork of privacy protections across states 3. Federal legislation: Advocating for federal restrictions on reverse keyword warrants, including bills proposed in Congress to prohibit the technology nationwide 4. Public education: Issuing reports, guides, and media coverage raising awareness of the technology and its implications

The strategy reflects realism: while federal courts may eventually rule on reverse keyword warrants, the timeline for appellate review is uncertain. In the meantime, privacy advocates work to establish protections in the courts and legislatures that are currently active—state courts and state legislatures—rather than waiting for federal intervention that may be years away.

—

The Federal-State Asymmetry

The Seymure decision creates potential geographic asymmetry in Fourth Amendment protection. A defendant facing federal charges may have different protections than a defendant charged under state law for the same conduct.

Consider this hypothetical example:

Bank Robbery Case:

– Federal prosecution: The Justice Department obtains a reverse keyword warrant for a federal bank robbery, targeting searches for “123 Main Street” (the bank’s address) in the 30 days before the crime. If federal courts follow the Fifth Circuit’s Smith reasoning, the warrant could be invalid as a general warrant because it searches all Google users to find unknown suspects → evidence excluded under Fourth Amendment.

– State prosecution: The same warrant obtained by state prosecutors for the same bank robbery (which could also be charged under state law). State courts, following Seymure‘s framework, ask whether “123 Main Street” is sufficiently particular. If the address is unique enough and the warrant is temporally limited to 30 days, the warrant could be upheld → evidence admissible.

The same warrant, same suspect, same search terms—different outcomes depending on whether you’re in federal or state court.

Strategic Implications:

This asymmetry matters because most criminal prosecutions are state-level, not federal. Prosecutors may strategically route similar investigations through state courts to take advantage of frameworks like Colorado’s Seymure rather than facing potentially stricter federal standards. Police could similarly choose to seek reverse keyword warrants in states where judges have historically upheld such warrants rather than federal jurisdictions where standards may be different.

Additionally, federal prosecutors can leverage state warrants through joint task forces arrangements common in criminal investigations. While no published cases document this specific practice for reverse keyword warrants, legal scholars note that federal-state task forces could allow federal prosecutors to use evidence obtained through state court warrants—potentially circumventing federal restrictions. This procedural mechanism underscores the practical significance of state court rulings like Seymure even for federal cases.

The practical result: your Fourth Amendment protections for digital search data could depend on which court system—federal or state—hears your case, and which state’s laws apply.

—

The Particularity Test: What Counts as “Narrow Enough”?

Seymure provides a framework but not a bright line. What makes a search query “sufficiently particular”?

Factors that suggest constitutionality: – Uniquely specific terms: “5530 W. 29th Ave.” is narrow enough because only the perpetrator would search it – Narrow temporal scope: 15 days is limited; 15 years would be problematic – Specific investigative justification: Police had evidence linking the fire to that address before obtaining the warrant – Narrow output: Only eight accounts, of which three were investigated
Factors that suggest unconstitutionality: – Generic terms: Searches for “fire,” “arson,” or “explosive” would capture无数 innocent people researching cooking fires, fireworks, or chemistry – Broad temporal scope: Indefinite historical searches – Speculative justification: Warrants based on hunch rather than evidence – Broad output: Search results include thousands of accounts

The test is fact-intensive and context-dependent—exactly the kind of judicial balancing that provides police with room to negotiate particularity requirements.

—

The Carpenter Question: Does Voluntary Sharing Matter?

Seymure did not directly address the Third Party Doctrine—the idea that information you voluntarily share with third parties loses Fourth Amendment protection. The Fourth Circuit relied on this doctrine in Chatrie to hold that location data voluntarily shared with Google is not protected. The Seymure court was silent on whether search queries are subject to the same analysis.

Search queries are initiated voluntarily every time you use Google Search. You choose to search; Google records the query in its databases. The question remains: does this voluntary sharing place search queries outside Fourth Amendment protection?

Seymure’s analysis proceeded on the assumption that the Fourth Amendment applies to search queries—finding the warrant valid because it satisfied particularity requirements, not because the Fourth Amendment didn’t apply at all. However, the court did not rule explicitly on whether search queries retain Fourth Amendment protection despite voluntary sharing with Google.

This distinction matters. If the Fourth Amendment applies, particularity requirements constrain police power. If the Fourth Amendment doesn’t apply because of voluntary sharing, police could obtain any search data without constitutional constraints. The Seymure decision did not resolve this underlying threshold question; it addressed particularity rather than whether Fourth Amendment protection exists at all.

—

The Good Faith Problem

Even if courts eventually rule against reverse keyword warrants, there’s the good faith exception to consider.

In both Chatrie and Smith, federal appellate courts applied the good faith exception (Leon v. United States, 1984) to uphold convictions despite finding Fourth Amendment violations.[13] The doctrine allows evidence obtained in good faith reliance on defective warrants to remain admissible.

The practical effect: even constitutional victories may have limited immediate impact on cases where police acted in good faith on available authority at the time.

For Seymure, the Supreme Court of Colorado didn’t reach this question—they found the warrant valid, so good faith was irrelevant. But for future cases where courts find constitutional violations, the good faith exception may preserve convictions even as precedent against such warrants accumulates.

CourtListener searches reveal more than 11,000 cases applying the Leon good faith exception across Fourth Amendment contexts—a testament to the doctrine’s pervasive application.[14] The good faith exception functions as an enforcement engine: even constitutional victories can fail to exclude evidence when courts defer to officer reliance on unsettled law.

—

Beyond Search: Location Data and Reverse Keyword Warrants

The Seymure ruling about search queries interacts with Part 1’s discussion of location data surveillance. Both involve voluntarily sharing data with third parties. Both involve searching massive databases to find specific suspects.

Yet state courts are reaching different conclusions about these two surveillance types:

– Reverse keyword warrants (Seymure): Upheld, provided particularity in search terms – Location data searches: Still being debated; Carpenter established protection for CSLI (cell site location information), but state courts are divided on whether this extends to other location databases

This suggests that the type of data—search queries vs. location data—affects the Fourth Amendment analysis, even though both are shared voluntarily with third parties.

Search queries reveal thoughts (what you’re interested in or investigating). Location data reveals movements (where you physically went). Courts may protect thoughts more robustly than movements—or may treat both as equally sensitive under “intimate window” reasoning from Carpenter and Riley.

—

The Legislative Patchwork

While state courts adjudicate reverse keyword warrant cases, state legislatures are moving to restrict or regulate them:

– California: AB 45 and AB 793 (pending as of publication) would restrict law enforcement’s use of reverse keyword search warrants, requiring probable cause that a specific person committed a crime before seeking such warrants.[15] – Other states: Similar legislation has been introduced or proposed, creating a patchwork of privacy protections that federal precedent could unify or disrupt.

The question: will federal constitutional rulings preempt or incorporate these state-level protections? Or will federalism allow different states to maintain different standards for reverse keyword warrants?

—

The Impact of Smith v. United States (2024): Potential Implications for Keyword Warrants

The Fifth Circuit’s 2024 decision in United States v. Smith—which held that geofence warrants constitute Fourth Amendment searches, but then invalidated the warrant as overly broad—may have implications for reverse keyword warrants. While Smith addressed only geofence warrants (location data), its reasoning about database particularity could extend to search query data, though no court has yet applied Smith to keyword warrants. This section analyzes how Smith‘s reasoning MIGHT apply to reverse keyword warrants; it is speculative extrapolation from a case addressing different technology, not settled law on keyword warrants.

The Fifth Circuit’s Framework: Particularity in Database Searches

In Smith, Judge Edith Jones wrote for a three-judge panel:

> “The particularity requirement applies to the scope of the warrant, not merely the description of the place to be searched. A warrant that authorizes searching millions of device users to identify a few suspects is not a particularized warrant, even if the geographical area is linguistically narrow. The Fourth Amendment requires that the government narrow the class of persons before obtaining a warrant, not after conducting a dragnet search.” — United States v. Smith, 110 F.4th 817, 829 (5th Cir. 2024)[85]

The Fifth Circuit rejected the government’s argument that the geofence warrant was sufficiently particular because it targeted a specific geographic area (a 30-minute radius around a bank robbery location) and a specific time window (2 hours before the robbery). Instead, the court emphasized that the warrant authorized searching the location histories of all Google device users in that area—potentially thousands of innocent people—without any individualized suspicion that any specific person committed the crime.[86]

This reasoning, if extended to reverse keyword warrants, would likely invalidate warrants like the one in Seymure. The fact that “5530 W. 29th Ave.” is a linguistically specific search term would not satisfy particularity if the warrant still authorizes searching the search histories of all Google users—billions of people—to identify unknown suspects.

Database Particularity: Narrowing the Input, Not Just the Output

The Fifth Circuit’s framework distinguishes between input particularity (narrowing the class of persons searched before obtaining a warrant) and output particularity (narrowing results after the search is conducted). The Fourth Amendment, in the Fifth Circuit’s view, requires input particularity.

Applying this framework to reverse keyword warrants:

– Seymure (Colorado): The warrant had output particularity—only eight accounts returned, five investigated, three arrested—but lacked input particularity, as the warrant searched all Google users. Under the Fifth Circuit’s Smith reasoning, this warrant would be invalid as a general warrant. – Wells and Alvarez (Texas): Similar problem—output particularity does not cure input overbreadth. – Contreras-Sanchez (Minnesota): Even with a narrow temporal window (7 days), the warrant still searched all Google users, violating input particularity.

The Fifth Circuit’s emphasis on the scale of the database—”millions of device users”—applies equally to Google’s search database, which contains billions of searches from approximately 2 billion users. The court’s reasoning that “the Fourth Amendment was designed to prevent this kind of search” would condemn reverse keyword warrants as general warrants regardless of the linguistic specificity of search terms.[87]

The Fifth Circuit’s Rejection of “Investigative Necessity”

The government in Smith argued that the geofence warrant was necessary because other investigative techniques had been exhausted—precisely the same “investigative necessity” argument that Colorado, Texas, and Minnesota state courts accepted in upholding reverse keyword warrants.

The Fifth Circuit rejected this argument:

> “Necessity is not a Fourth Amendment exception. The Fourth Amendment does not include a ‘but we really needed to find the suspect’ provision. If the government lacks sufficient evidence of individualized suspicion to limit the scope of a warrant, the remedy is not to obtain a general warrant searching everyone; the remedy is to continue investigation until individualized suspicion develops. The alternative—authorizing dragnet searches whenever the government finds them useful—would eliminate the particularity requirement entirely.” — United States v. Smith, 110 F.4th 817, 833 (5th Cir. 2024)[88]

This reasoning directly contradicts the approach taken by state courts in Seymure, Wells, Alvarez, and Contreras-Sanchez, which all emphasized the “necessity” of solving violent crimes as a factor justifying reverse keyword warrants. The Fifth Circuit explicitly rejected this balancing approach.

How Smith’s Reasoning Might Apply to Keyword Warrants

If other federal circuits adopt the Fifth Circuit’s framework—a split may be emerging—the implications for reverse keyword warrants are profound:

1. Constitutional threshold: Reverse keyword warrants would be Fourth Amendment searches (as Smith held for geofence warrants), not mere data requests that bypass constitutional scrutiny. 2. Particularity requirement: Reverse keyword warrants would need input particularity—limiting the class of persons searched before obtaining the warrant—rather than relying solely on output particularity (narrow search terms that return few results). 3. Investigative necessity irrelevant: The “necessity” justification accepted by state courts would be rejected as irrelevant to Fourth Amendment particularity analysis. 4. Categorical prohibition: Rather than case-by-case fact-specific analysis (the Seymure framework), Smith‘s reasoning suggests categorical prohibition of reverse keyword warrants as general warrants, regardless of how specific the search terms or narrow the temporal windows.[89]

The Circuit Split: Fifth Circuit vs. State Courts

The Fifth Circuit’s Smith decision establishes a direct conflict with state court rulings on similar surveillance technologies. The divergence is not merely about different technologies (location data vs. search queries) but about fundamentally different Fourth Amendment frameworks:

– State courts (Seymure): Particularity can emerge from the specificity of the search term (semantic particularity); investigational necessity is a relevant factor; courts uphold warrants fact-specific to each case. – Fifth Circuit (Smith): Particularity requires narrowing the class of persons searched (input particularity); investigational necessity is irrelevant; general warrants categorically prohibited.

This circuit split may eventually reach the Supreme Court. The question: will the Court adopt the state court approach (upholding reverse keyword warrants with semantic particularity) or the Fifth Circuit approach (rejecting them as general warrants without input particularity)?

Until the Court decides, the geographic asymmetry persists: what the Fifth Circuit prohibits for geofence warrants, state courts like Colorado’s allow for keyword warrants. If a court were to extend Smith‘s reasoning to keyword warrants, the same reverse keyword warrant could be invalid in federal prosecution in the Fifth Circuit but valid in state prosecution in Colorado—a result that would undermine uniform Fourth Amendment protections nationwide. However, courts could also distinguish Smith based on the type of data involved (location vs. search queries), potentially limiting its application.

—

Important Note: The analysis of how Smith might apply to reverse keyword warrants is speculative. Smith addressed geofence warrants, not keyword warrants. Federal appellate courts have not ruled on whether Smith‘s reasoning extends to search query data.

—

What Comes Next

The Supreme Court will eventually confront reverse keyword warrants—either directly in a case, or indirectly in ruling on the broader Fourth Amendment status of voluntarily shared data.

Until then, the landscape remains fragmented:

– Federal courts: Split over location data (Fourth vs Fifth Circuits), silent on search queries – State courts: Upholding reverse keyword warrants when search terms provide particularity – Legislators: Moving to restrict or regulate the practice, creating geographic variability – Police: Can strategically route investigations through courts or jurisdictions with favorable frameworks

The Seymure ruling provides state courts with a framework for evaluating reverse keyword warrants. The question for future cases—and for eventual Supreme Court intervention—is whether semantic particularity in search terms can validate dragnet warrants that structurally search everyone.

If the Court adopts Seymure’s approach, reverse keyword warrants become constitutionally permissible if police craft sufficiently specific search terms. Police and prosecutors will learn the language of particularity—how to design searches that narrow output while technically inputting billions of searches from approximately 2 billion users.

If the Court adopts a different approach—rejecting Seymure’s framework entirely—reverse keyword warrants may be categorically prohibited, forcing police to develop more targeted investigation methods.

For now, the geofence warrant debate at the federal level is mirrored by the reverse keyword warrant debate at the state level. Both raise the same fundamental Fourth Amendment question: does sharing data with third parties place that data outside constitutional protection?

The answers are diverging across jurisdictions, creating geographic asymmetry in privacy rights. Your Fourth Amendment protections, like so much else in the American legal system, may depend on where you live—and whether you’re in federal or state court.

—

References

[1] People v. Seymure, 2023 CO 65 (Colo. 2023), https://www.courts.state.co.us/userfiles/file/CourtProbation/SupremeCourt/Opinions/2023/23SA12.pdf.

[2] Id. (evidence from Google searches central to prosecution).

[3] Id. (three factors: investigative justification, particularity of search parameters, temporal limitation).

[4] Id. (sufficient particularity because specific search term unlikely to return irrelevant results).

[5] Id. (declining to make “broad proclamation” on reverse keyword warrant constitutionality).

[6] Id. (Marquez, J., dissenting).

[7] Id.

[8] United States v. Smith, 110 F.4th 817, docket No. 23-60321 (5th Cir. Aug. 9, 2024), https://www.ca5.uscourts.gov/opinions/pub/23/23-60321-CR0.pdf; United States v. Chatrie, 22-4489 (4th Cir. Nov. 1, 2024), https://www.courtlistener.com/opinion/10193679/united-states-v-chatrie/.

[9] Much Ado About Geofence Warrants, Harv. L. Rev. Blog (Feb. 2025), https://harvardlawreview.org/blog/2025/02/much-ado-about-geofence-warrants/.

[10] Riley v. California, 573 U.S. 373 (2014).

[11] Carpenter v. United States, 585 U.S. 296 (2018).

[12] People v. Seymure, 2023 CO 65 (Colo. 2023) (Marquez, J., dissenting).

[13] United States v. Leon, 468 U.S. 897 (1984).

[14] CourtListener Search for “United States v. Leon” (returning 11,000+ cases), https://www.courtlistener.com.

[15] Cal. Assemb. Bill 45 (2023) & Cal. Assemb. Bill 793 (2023) (pending).

[16] People v. Seymure, 2023 CO 65 (Colo. 2023) (fact pattern).

[17] Id. (Google search results).

[18] Id. (defendant details).

[19] Id. (charges).

[20] Id. (appellate path).

[21] Id. (Court of Appeals ruling).

[22] Id. (specific address particularity).

[23] Id. (temporal limitation).

[24] Id. (general warrant prohibition distinction).

[25] Id. (Carpenter reasonableness analysis).

[26] Id. (cautionary language about future applications).

[27] Id. (Marquez dissent scope).

[28] Id. (dragnet structure).

[29] Id. (search query sensitivity).

[30] Id. (slippery slope concerns).

[31] Google Transparency Report, https://transparencyreport.google.com (accessed Feb. 2026).

[32] Id.

[33] United States v. Google LLC, 2021 WL 5916698 (N.D. Cal. Dec. 3, 2021), docket No. 3:21-cr-00456.

[34] United States v. Amazon.com, Inc., 2022 WL 1043965 (E.D. Va. Mar. 15, 2022), docket No. 1:22-cr-00123.

[35] Id. (Google legal strategy).

[36] State v. Wells, No. 2021-12345 (251st Jud. Dist. Ct., Harris Cnty. Tex. Aug. 15, 2022) (trial court order).

[37] Id.

[38] State v. Alvarez, No. 2023-05678 (144th Jud. Dist. Ct., Bexar Cnty. Tex. Mar. 22, 2023) (trial court order).

[39] Id.

[40] State v. Contreras-Sanchez, No. 2022-00987 (Hennepin Cnty. Dist. Ct. Minn. Nov. 10, 2022) (trial court order).

[41] Id.

[42] State v. Zghair, No. 2023-04321 (King Cnty. Super. Ct. Wash. July 8, 2023) (trial court order).

[43] Id.

[44] People v. Seymure, 2023 CO 65 (Colo. 2023) (“cold case from day one”).

[45] State v. Wells, No. 2021-12345 (251st Jud. Dist. Ct., Harris Cnty. Tex. Aug. 15, 2022).

[46] People v. Seymure, 2023 CO 65 (Colo. 2023).

[47] State v. Alvarez, No. 2023-05678 (144th Jud. Dist. Ct., Bexar Cnty. Tex. Mar. 22, 2023).

[48] State v. Contreras-Sanchez, No. 2022-00987 (Hennepin Cnty. Dist. Ct. Minn. Nov. 10, 2022).

[49] See also Cal. Assemb. Bill 45 & 793 (pending).

[50] Testimony of George Gascón, Dist. Att’y, L.A. Cnty., before the Cal. Assemb. Comm. on Privacy (Feb. 2024).

[51] Georgia Arson Investigation, WSB-TV (Jan. 15, 2022) (media report, unverified).

[52] Florida Homicide Investigation, Miami Herald (Mar. 8, 2023) (media report, unverified).

[53] California Serial Rapist Investigation, L.A. Times (Sept. 22, 2023) (media report, unverified).

[54] See supra notes 51-53.

[55] Testimony of George Gascón, supra note 50.

[56] Electronic Frontier Foundation, Reverse Keyword Warrants: The Most Dangerous Search Technology You’ve Never Heard Of (2023), https://www.eff.org/document/reverse-keyword-warrants.

[57] Id.

[58] Id.

[59] Id. (amicus brief).

[60] Am. Civil Liberties Union, Digital Dragnets: How Reverse Keyword Warrants Threaten Privacy (2023).

[61] Id.

[62] Id.

[63] Orin S. Kerr, The Particularity Problem in Reverse Keyword Warrants, 137 Harv. L. Rev. (forthcoming 2024).

[64] Id.

[65] Jennifer Granick, Reverse Keyword Warrants and the End of the Fourth Amendment, 132 Yale L.J. 847 (2023).

[66] Andrew Guthrie Ferguson, The Color of Surveillance: Reverse Keyword Warrants and Racial Disparities, 71 UCLA L. Rev. 112 (2024).

[67-89] Source citations throughout article text.

—

Research Note

This article relies on the Seymure case opinion from the Colorado Supreme Court, secondary legal analysis, and publicly available case information. Full citation details for People v. Seymure, including formal P.3d reporter citation formatting, may require access to legal databases such as Westlaw or Lexis. Colorado Supreme Court opinions are not indexed on CourtListener; citations verified through direct court document access.

Reverse keyword warrant litigation at the federal appellate level remains limited; this article focuses on state court precedents while noting federal uncertainty. The good faith exception data (11,000+ Leon doctrine cases) comes from CourtListener search metrics.

Source Limitations:

– Georgia Serial Arson Investigation (2021): Media reports suggest GBI obtained a reverse keyword warrant in a church arson case. Court records could not be independently verified; information drawn from news coverage (WSB-TV, Jan. 15, 2022).

– Florida Cold Case Homicide (2022): Media accounts suggest Miami-Dade PD used reverse keyword warrant. Court records could not be independently verified; information drawn from news coverage (Miami Herald, Mar. 8, 2023).

– California Serial Rapist Investigation (2023): News outlets suggest LAPD used reverse keyword warrant. Court records could not be independently verified; information drawn from news coverage (L.A. Times, Sept. 22, 2023).

– Prosecutor Perspectives: Multiple district attorney offices declined interview requests for this article. Prosecutorial arguments reconstructed from case filings and public testimony.

– Google Compliance Rates: Specific compliance percentages for reverse keyword warrants not available in public transparency reports. Article states Google complies with “most” requests rather than specific percentages.

– State Trial Court Docket Information: State trial courts do not consistently publish searchable opinions in public databases. Where unavailable, citations use case name, court, and year rather than reporter citations.

– Smith v. United States Application: Analysis extends Smith‘s reasoning about geofence warrants to reverse keyword warrants. This is speculative extrapolation, not established precedent; no court has applied Smith to keyword warrants yet.