In Chatrie v. United States, a divided Supreme Court held that police conduct a Fourth Amendment “search” when they obtain a person’s cell-phone location data from Google through a geofence warrant. Writing for a five-Justice majority, Justice Kagan concluded that “an individual has a reasonable expectation of privacy in his cell-phone location information,” and that this protection holds even when the data covers only a short time window and even when it is handed over by a third-party technology company.
The ruling is a direct descendant of Carpenter v. United States (2018), and the majority leaned on that precedent so heavily that even the dissent accused it of “rely[ing] primarily” on Carpenter rather than older doctrine – a charge the majority answered by pleading “guilty as charged.” But the decision did not end the case. The Court resolved only whether a search occurred, leaving the separate question of whether the unusual multi-step warrant was reasonable — that is, whether it satisfied the Fourth Amendment’s probable-cause and particularity requirements — for the Fourth Circuit to address on remand.
Before getting into the opinion itself, it’s worth understanding the technology at the center of the case and, for anyone facing charges, how to tell whether one of these warrants was used against you.
What Is a Geofence Warrant?
What a geofence warrant actually is
A geofence warrant is a court order that works backward from a crime to a suspect. Instead of naming a person and asking for their data, police draw a virtual perimeter — the “geofence” — around a location, pick a window of time, and compel a company (almost always Google) to hand over data on every device that was inside that box during that window. Because it starts with a place and time rather than a named target, it’s often called a “reverse” warrant.
The whole point is to identify an unknown suspect. As one court put it, the goal is to find out “who was there and so who might have done it.” Geofence requests typically run through a three-step funnel: an anonymized list of all devices, then expanded movement data for a narrowed subset, then real names and account details for the final few. (That exact process is described in detail in the case summary below.)
How that differs from “regular” cell-phone location data
This is the distinction that confuses most people, so here it is side by side.
| Targeted location data (the “ordinary” kind) | Geofence warrant (the “reverse” kind) | |
|---|---|---|
| Starting point | A known person or phone — police already have a suspect | A place and time — police have no suspect yet |
| The question asked | “Where did this phone go?” | “Whose phones were here?” |
| Who gets swept in | Just the target | Everyone in the area — suspects, witnesses, and unrelated bystanders alike |
| Typical source | Cell-site location info (CSLI) from a wireless carrier (AT&T, Verizon, T-Mobile), or a phone’s own GPS | Google “Location History” / the Sensorvault database (and sometimes Apple, Lyft, Snapchat, Uber) |
| How precise | CSLI is coarse — roughly an eighth of a mile to several miles | Location History is fine — within about 20 meters, sometimes down to the floor of a building |
| Legal label | Governed by Carpenter v. United States (2018) for CSLI | Now governed by Chatrie v. United States (2026) |
In short: a traditional location request is a spotlight aimed at one person. A geofence warrant is a dragnet cast over a location. After Chatrie, both require a warrant, but the dragnet raises a second set of problems (probable cause and particularity at each step) that the Supreme Court left for the lower courts to sort out.
One important practical wrinkle
In 2023–2025, Google changed how Location History works — storing the data on a user’s own phone rather than on Google’s central servers, and shortening how long it’s kept. Google now says it generally cannot respond to these warrants for newer data. That doesn’t help anyone whose case predates the change, and it doesn’t affect carrier-based CSLI or other companies’ data — so geofence evidence will keep surfacing in older and pending cases for years.
How would you know a geofence warrant was used in your case?
This is the hard part, because a geofence warrant often doesn’t have your name on it. You were “Device 7” on an anonymized list before you were ever a suspect. Police may then build a parallel record — describing how they “developed” you as a lead — that obscures the geofence as the true starting point. Here is where to look.
- Read the charging documents for a vague origin story. If the affidavit or police report says investigators identified you through “investigative means,” “information from a third party,” “data analysis,” or “a tip” without explaining the actual first step, that gap is a red flag worth running down.
- File for full discovery — and ask specifically. Don’t rely on a general discovery demand. Counsel should request, by name, any search warrants, applications, returns, and supporting affidavits directed to Google (or Apple, etc.), plus any “reverse location,” “reverse keyword,” or “tower dump” requests. The defense team in Chatrie filed a dedicated discovery motion aimed squarely at Google’s Sensorvault data; that motion is a public template.
- Look for the anonymized-ID paper trail. The geofence process produces spreadsheets of devices tagged with temporary anonymous identifiers (Google assigns a per-warrant device ID; carriers may use IMSI numbers). If discovery includes a list of numbered or coded “devices” with timestamps and coordinates, you are almost certainly looking at a geofence return.
- Check for delayed-notice or sealing orders. These warrants are frequently sealed, and companies are often gagged from notifying users. A motion to unseal warrant materials may be necessary. Some states (for example, California under CalECPA) require eventual notice to the target — so absence of notice isn’t proof one wasn’t used.
- Subpoena or request records directly from the provider. Google maintains a Law Enforcement Request System and can confirm what was produced about a given account. Counsel can also seek the provider’s declaration describing exactly what was searched and returned (a declaration of the type used in Chatrie).
What language to look for in the warrant or affidavit
Geofence warrants share a recognizable vocabulary. If you see these phrases in a warrant, application, or supporting affidavit, you are very likely looking at a geofence (or a close cousin like a tower dump or keyword warrant).
Tell-tale phrases:
- “Devices located inside the geographical region(s) bounded by the following latitude / longitude coordinates” — the defining language of a geofence.
- “Google Location History,” “Sensorvault,” or “the Google Account(s) associated with devices” in a given area.
- References to an “initial search area” or a search defined by a radius from a point or a polygon, plus a specific date and time window.
- A staged or “multi-step” production protocol — language about producing “anonymized” data first, then “contextual data points” or movement “outside the geographical area” for a subset, then “identifying account information / subscriber information” (user name, date of birth, email addresses, telephone numbers, devices associated with the account).
- “Anonymized device identifier,” “obfuscated ID,” or a reference to expanding the time frame “30 minutes before and 30 minutes after” an initial window (the exact phrasing used in the Chatrie warrant).
- Boilerplate justifying the technique: statements that “most people carry cellular phones on their person,” that “suspects involved in criminal activity will typically use cellular phones to communicate,” and that Google “tracks the location of devices that use at least one Google application … regardless of Android or iOS.”
Related warrants that use similar language:
- Tower dump: asks a carrier for all phones that connected to a specific cell tower during a window — same dragnet logic, carrier data instead of Google.
- Reverse keyword warrant: asks Google for everyone who searched a particular term — look for “search query,” “keyword,” or “users who searched for.”
What makes one vulnerable to challenge: Even after Chatrie confirmed these are searches, the warrant can still fail on its own terms. The Chatrie trial judge faulted the original affidavit as barely a page long, with no explanation of how the geofence would separate the guilty from innocent bystanders. Justice Jackson’s concurrence zeroed in on language at the narrowing stages that only said officers would “attempt to narrow down the list” — with no criteria and no return trip to a judge. When reviewing a geofence warrant, the questions are whether it was particular in time, location, and scope, and whether a magistrate — not an officer’s later discretion — actually authorized each step.
Bottom line: If your case involves an unsolved crime that police suddenly “solved,” a vague account of how you became a suspect, and any reference to Google, location coordinates, anonymized devices, or a staged data hand-off, ask directly about a geofence warrant. The single most useful step is a targeted discovery demand naming the provider and the words above.
Back to the Case: A Bank Robbery and a Virtual Perimeter
On May 20, 2019, at around 4:50 p.m., a man robbed a credit union in Midlothian, Virginia. He handed a teller a note demanding $100,000, threatened her and her family, claimed to have “boys on the lookout outside,” brandished a firearm, and forced the manager to load roughly $195,000 into a bag before fleeing on foot.
Local police learned from witness interviews and surveillance footage that the robber had approached the credit union from a corner of an adjacent church while appearing to talk on a cell phone — but the trail otherwise went cold. On June 14, officers applied to a Virginia magistrate for a geofence warrant directed at Google.
The warrant targeted Location History, a Google service that records a cell phone’s position roughly every two minutes, drawing on Wi-Fi, Bluetooth beacons, cell sites, GPS, and IP data. When combined, those signals can pinpoint a phone within about 20 meters and can even estimate elevation — revealing which floor of a building a phone is on. At the time of the warrant, over 500 million users worldwide had enabled Location History, and Google stored that data in a central repository on its own servers.
The three-step protocol
The warrant followed a three-step process Google had developed with law enforcement:
- Step one: Google produced anonymized location data for all phones within the 150-meter geofence during the hour from 4:20 to 5:20 p.m. (30 minutes before to 30 minutes after the robbery). This yielded 19 users.
- Step two: Officers narrowed the list and Google supplied additional anonymized data, both inside and outside the geofence, over an expanded two-hour window (3:50 to 5:50 p.m.). The list was cut to 9 users.
- Step three: Officers narrowed again, and Google turned over identifying information (names, email addresses, phone numbers) for the final list of 3 users.
One of those three was Okello Chatrie, whose data showed he entered the geofenced area about ten minutes before the robbery and headed toward a residential neighborhood immediately afterward. A federal grand jury later charged him with robbery and related firearms offenses.
The Procedural Journey
Chatrie moved to suppress the Google data, arguing the officers had conducted a Fourth Amendment search under an invalid warrant. The lower courts splintered:
- District Court (E.D. Va.): Found that the geofence warrant “plainly violates the rights enshrined in [the Fourth] Amendment,” but denied suppression under the good-faith exception to the exclusionary rule (per United States v. Leon).
- Fourth Circuit panel: Affirmed on different reasoning, holding that no search occurred because Chatrie had no reasonable expectation of privacy in two hours of Location History “voluntarily exposed to Google.” Judge Wynn dissented.
- Fourth Circuit en banc: Affirmed in a one-sentence per curiam, with the court dividing evenly (7–7) on whether a search had occurred. Of the seven who thought one had, most believed the warrant defective — but most also thought the good-faith exception applied, so they ruled against Chatrie anyway.
The Supreme Court granted certiorari solely on the search question, expressly declining to take up the exclusionary-rule issue. That choice — to answer the search question while leaving the good-faith ground untouched — became the centerpiece of Justice Alito’s dissent.
The Majority Opinion (Justice Kagan)
Joined by Chief Justice Roberts and Justices Sotomayor, Kavanaugh, and Jackson.
The majority framed the case as the latest in a line of decisions adapting Fourth Amendment principles to new technology — from Riley v. California (cell phones searched incident to arrest) to Kyllo (thermal imaging) to Carpenter itself. The Amendment’s “basic purpose,” the Court reiterated, is “to place obstacles in the way of a too permeating police surveillance.”
Location History as the heir to Carpenter
Carpenter held that accessing cell-site location information (CSLI) is a search because “individuals have a reasonable expectation of privacy in the whole of their physical movements.” The majority found that “the resemblances between CSLI and Location History… practically leap off the page,” and that everything Carpenter relied on “applies as well or better” to Location History on three fronts:
- Precision. CSLI placed a suspect within a sector of one-eighth to four square miles; Location History pinpoints location to roughly 20 meters — less than 2% of a mile — and logs a position about every two minutes (a daily average of 720 chartings versus CSLI’s 101). It can even reveal a building’s floor.
- Retrospective reach. Like CSLI, Location History lets police reconstruct movements “with no real effort,” enabling “tireless and absolute surveillance” of any number of people in any number of places, at “the click of a button.”
- Personal ownership. Location History “implicate[s] those privacy interests still more than CSLI, because the former is more the individual’s own.” Where most users have no awareness of CSLI records, Google users treat Location History as a personal journal — consulting it to recall a restaurant or a friend’s home. In that respect it resembles emails, photographs, or calendars, which a user “reasonably views as his own” even when stored on Google’s servers.
Rejecting the “short duration” argument
The Government’s central position was that two hours of data is too brief to count as a search. The Court rejected this on several grounds:
- Even “short-term monitoring” can reveal “a wealth of detail about [a person’s] familial, political, professional, religious, and sexual associations” (quoting Justice Sotomayor’s Jones concurrence) — trips to a psychiatrist, an abortion clinic, an AIDS treatment center, a political rally.
- The Fourth Amendment has never been understood to “kick in only once an intrusion goes too far.” Where it applies, it applies “regardless of the quality or quantity of information” obtained (citing Kyllo). The Court analogized to its seminal wiretap case, Katz, where police captured only 18 minutes of recordings.
- A “grace period” approach would breed unanswerable line-drawing problems: Is the cutoff two hours? Six? One day? Does the clock reset? Could two parallel investigations double the permissible access?
- Crucially, when officials “can select the time-limited set of materials they want from an all-encompassing database,” the short duration is “more a practical benefit to the government than a limit on its intrusive powers.”
The Court also distinguished United States v. Knotts (the 1983 beeper case), noting that Knotts expressly cabined itself to “rudimentary” technology and involved surveillance confined to public roads — unlike Location History, which “faithfully follows” a phone into private residences.
Rejecting the third-party doctrine
The Government’s fallback was the third-party doctrine — the rule from United States v. Miller (bank records) and Smith v. Maryland (dialed phone numbers) that a person loses Fourth Amendment protection in information voluntarily conveyed to others. But Carpenter had already refused to apply that doctrine to CSLI, on two grounds that the majority found applied “equally or better” to Location History:
- It is qualitatively revealing. There is “a world of difference” between an exhaustive chronicle of movements and the limited records in Smith and Miller.
- It is not “truly shared.” Disclosure to Google “is merely what happens when a user avails himself of one of the services on his cell phone” — the automatic price of ordinary cell-phone use.
The Government argued that, unlike CSLI, Location History is a voluntary “optional add-on” — noting that only about one-third of Google accountholders enable it. The Court was unpersuaded:
- Google “repeatedly prompts” users to turn the service on — at account setup, app setup, and phone setup — often warning Android users their devices won’t “work correctly” otherwise, while not disclosing how often location is recorded, how precise it is, or that it might be handed to the government.
- The one-third figure is “almost surely overstated,” since it appears to include users in countries like China where collecting Location History is illegal.
- A “feature-by-feature method of granting Fourth Amendment protection misapprehends the very nature of modern cell-phone use,” where “[p]retty much everything a person does on a smartphone requires some kind of opt-in.” The Government’s logic would treat us all “as living in dumb flip-phone days.”
The Court therefore held that police invade a reasonable expectation of privacy when they access Location History — “It does not matter if the time period scrutinized was only two hours. Nor does it matter that the materials… were handed over by a third-party tech company.”
What the Court left open
Finding a search did not end the case. The Fourth Amendment bars only unreasonable searches, and a warrant is generally required from “a neutral and detached magistrate,” issued only on probable cause with a particularly described scope. The warrant here was “an uncommon, multi-step one,” and the parties contested every stage. Because the Fourth Circuit had never reached those questions, the Court — “a court of review, not of first view” — vacated and remanded for the Court of Appeals to decide whether each step satisfied probable cause and particularity.
The majority also noted (in a footnote) that its decision does not disturb the possibility of a warrantless geofence search under exigent circumstances, mirroring Carpenter‘s treatment of CSLI.
A technological coda: In July 2025 — years after this warrant — Google changed Location History to store data on users’ individual devices rather than on its own servers. As a result, Google represents that it can no longer respond to geofence warrants seeking that data, a fact the dissent seized on to argue the procedure is now “obsolete.”
The Concurrences
Justice Jackson (joined by Justice Sotomayor)
Justice Jackson would have gone further and held the search unconstitutional now. In her view, steps two and three plainly failed the probable-cause and particularity requirements: the warrant only said officers would “attempt to narrow down the list,” set out no criteria for doing so, and let officers gather sensitive data — ultimately revealing trips to residences, a school, and a hospital — without ever returning to a magistrate. This gave officers a “roving commission” (quoting Berger v. New York), and the list was narrowed at step one only because Google insisted on it, not because the warrant required it. She urged the Fourth Circuit to keep this in mind on remand.
Justice Gorsuch (concurring in the judgment)
Justice Gorsuch agreed there was a search but rejected the Katz “reasonable expectation of privacy” framework entirely, calling it textually and historically unmoored, unworkable, and — together with its “battered third party doctrine” — an exercise in “we know it when we see it.” He would instead return to the Fourth Amendment’s text, asking whether Location History is one of Chatrie’s “papers” or “effects.”
His answer: it is an “effect,” meaning personal property. Chatrie could review, edit, export, and delete the data; Google’s own agreement called it “your information” and promised to protect it. Citing state computer-crime statutes (Virginia, Texas, Georgia and others) and cases treating digital data as property, Gorsuch reasoned that Chatrie held the key “sticks in the bundle of rights” — including the “most treasured” right to exclude. Entrusting data to Google no more forfeits ownership than tossing keys to a valet or leaving a dog with a neighbor. He found “hints” of this property-based reasoning lurking within the majority’s own opinion.
The Dissents
Justice Alito (joined in part by Justices Thomas and Barrett)
Justice Alito’s lengthy dissent advanced two principal lines of attack.
Part I — The opinion is advisory. Alito argued the Court should have dismissed the case or affirmed on good-faith grounds. Because the Fourth Circuit’s judgment rested independently on the good-faith exception — and because the majority’s opinion “does not disturb” that basis — “not one iota” of the decision affects the outcome for Chatrie. He acknowledged the Court technically had Article III jurisdiction (the conviction made it a live case), but invoked the Court’s longstanding prudential policy against gratuitous constitutional pronouncements. The majority’s grant of certiorari that excluded the good-faith question, he charged, “carefully set the stage for its planned performance: striking a pose as a great champion of privacy in the digital age.” He also noted the procedure is now obsolete, making the case a poor vehicle. (Justice Thomas joined this part.)
Part II — The merits are wrong. Alito argued that under both 19th-century principles (document-production orders were not “searches,” and the Amendment protected only one’s own papers and effects) and 20th-century doctrine (the third-party doctrine of Miller and Smith), there was no search. Even under Carpenter, he contended, three factors cut decisively against Chatrie:
- Duration — two hours, versus the 127 days of data in Carpenter (Alito repeatedly invoked the 127-day figure; the majority countered that Carpenter‘s actual holding drew the line at seven days).
- Comprehensiveness — the geofence centered on a credit union, a public place, not a chronicle of someone’s every movement.
- Voluntariness — Location History is genuinely optional and not integral to a phone’s function, unlike the unavoidable generation of CSLI.
He warned that the majority’s rule — requiring a warrant for any third-party cell-phone location data, “however brief… however innocuous… however voluntarily” disclosed — “unshackles” Carpenter and will “unleash the very upheaval” Carpenter disclaimed. He pressed a battery of line-drawing questions the majority left unanswered: Do police now need warrants for Amazon purchase histories? Google searches? Venmo logs? Apple Pay data? The majority’s “location information” qualifier, he predicted, “might as well be written on the dissolving paper sold in magic shops.”
Justice Barrett
In a brief solo dissent, Justice Barrett distanced herself from Alito’s broader assault. She had “no quarrel with Carpenter” and no objection to granting certiorari. But she agreed that under existing precedent — Carpenter included — Chatrie had no reasonable expectation of privacy in data about his public movements that he voluntarily disclosed to Google. She therefore respectfully dissented. (Barrett also joined the specific merits subsections of Alito’s dissent dealing with the third-party doctrine and the Carpenter factor analysis.)
Why the Decision Matters
A doctrinal expansion of Carpenter. The most consequential move is the majority’s rejection of any duration threshold. Carpenter expressly reserved whether some “limited period” of location data might be obtainable without a warrant. Chatrie answers no — for location data, the Fourth Amendment “applies regardless of the quality or quantity of information.” That untethers Carpenter from the seven-day line that had given lower courts a workable boundary.
The third-party doctrine continues to erode. By extending Carpenter‘s “not truly shared” reasoning to an admittedly optional service, and by reframing nearly all smartphone activity as involuntary in the relevant sense, the majority signals that Miller and Smith may have little purchase over modern digital records — even as it formally leaves them intact. Alito’s unanswered questions about purchase histories and search logs preview the next generation of litigation.
An unsettled remedy. Because the Court left probable cause, particularity, and good faith for the Fourth Circuit, Chatrie establishes that geofence searches require a valid warrant without yet specifying what a valid geofence warrant looks like. Justice Jackson’s concurrence sketches the likely battleground: the discretion these multi-step protocols hand to officers at the narrowing stages.
A possibly moot mechanism. The practical sting is softened by Google’s 2025 architecture change, which apparently removes its ability to answer these warrants at all. The decision’s lasting force lies less in the specific geofence procedure than in its broad principle: when the government taps a company’s “database of physical location information,” the Fourth Amendment applies — and, as Justice Kagan closed by quoting Carpenter and Justice Brandeis’s Olmstead dissent, the courts remain “obligated… to ensure that the progress of science does not erode Fourth Amendment protections.”
The Lineup at a Glance
| Justice | Position |
|---|---|
| Kagan (author) | Majority — a search occurred; remand on reasonableness |
| Roberts, Sotomayor, Kavanaugh, Jackson | Joined the majority |
| Jackson (joined by Sotomayor) | Concurrence — would hold steps two and three unconstitutional now |
| Gorsuch | Concurrence in judgment — search occurred, but via property/”effects” analysis, not Katz |
| Alito (Thomas joins Part I; Barrett joins Parts II-B, II-C-1, II-C-2) | Dissent — opinion is advisory and wrong on the merits |
| Barrett | Dissent — no expectation of privacy in voluntarily disclosed public movements |
Read the Full Decision
Download the slip opinion (PDF)
