GOOGLE, SENSORVAULT, AND THE SUPREME COURT
Ronald J. Rychlak
For the last several years, “geofence warrants” have been the cutting edge of digital policing. Investigators who knew where and when a crime occurred, but had no idea who did it, could go to Google and ask: “Show us every device that was here, at this time.” Google’s answer came from a massive repository called Sensorvault, which stored Location History Data (LHD) for millions of users. If you carried an Android phone or used Google apps with Location History on, your device’s comings and goings were almost certainly in that database.
From there, the process followed a familiar script. First, Google would return an anonymized list of devices that fell within a virtual perimeter, the geofence, around the crime scene over a particular time period. Next, police would ask for a bit more information about a subset of those devices: where they were, say, thirty or sixty minutes before and after the incident, to see who arrived, lingered, and left in a suspicious pattern. Finally, after winnowing the list further, officers would request names and account information for the most promising devices. That last step, de‑anonymizing the data, would turn dots on a map into people.
This three‑step protocol made geofence investigations tremendously effective. It also made them deeply controversial. Civil libertarians saw a digital dragnet: a single search that not only scooped up data on everyone near a particular scene but also technically searched through the records of anyone whose data was stored by Google, regardless of whether they were anywhere near the vicinity of the crime.
Police officials usually obtained a warrant somewhere along the way, most typically before the first request to Google, but sometimes at step three, at the time of de‑anonymizing the data. In an article published in the Mississippi Law Journal, I argued that the warrant was actually needed only at that time.[1]
Last year, the U.S. Court of Appeals for the Fifth Circuit, in a case called United States v. Smith, labeled geofence warrants “general warrants” and declared them categorically forbidden by the Fourth Amendment.[2] Others, like In re Search of Information Stored at Premises Controlled by Google, Washington, D.C., from the U.S. District Court for the District of Columbia; United States v. Rhine from the U.S. District Court for the District of Arizona; and the Texas Court of Criminal Appeals in Wells v. State, upheld tightly drawn geofences as constitutional when limited in time and space and when executed through Google’s anonymized, stepwise process.[3]
That clash is now headed to the Supreme Court in Chatrie v. United States, a case arising from a bank robbery in Virginia.[4] There, police used a geofence warrant to identify devices near the bank, tracked movements as they approached and left the scene, and ultimately unmasked an account linked to the defendant, Okello Chatrie.[5] The trial judge found the warrant defective but refused to suppress the evidence under the good‑faith exception.[6] A three-judge panel from the Fourth Circuit resulted in a two-to-one decision agreeing with the district court that the evidence should not be suppressed—but for a different reason: that Chatrie did not have a reasonable expectation of privacy in two hours’ worth of LHD that he had voluntarily exposed to Google.[7] Following an “en banc” re-hearing, the court issued an unsigned, one-sentence opinion joined by fourteen of the fifteen judges and affirming the district court’s decision to deny suppression of the geofence evidence, but with eight concurring opinions and one dissent.[8] In January 2026, the Supreme Court granted certiorari.[9]
As all of this was unfolding, Google made a seismic announcement: it would no longer keep users’ LHD in a central server‑side database.[10] Instead, from now on, LHD will be kept on users’ own devices. This change matters for three main reasons.
First, it undercuts the technical premise that made geofence warrants so attractive. The original model depended on the fact that there was one easily described place to be searched: Google’s servers. There was also one easily described thing to be seized: LHD from within the geofence. That made it possible to write a warrant that looked like most others and seemed to meet the constitutional requirements. With data now stored on each device, there is no single third‑party vault to query. To get LHD, police need to know whose phone to search, and they don’t. Geofence warrants worked well when there was a known crime scene but an unknown suspect. That’s no longer true; the tool no longer fits the job.
Second, the shift changes the shape of the legal debate. Many of the fiercest objections to geofence warrants focused on the idea of mass, suspicionless searching of everyone’s movements in a given area. The Fifth Circuit’s Smith decision leaned hard on history, arguing that a warrant authorizing such a broad sweep—before the government has any individualized suspicion—looks perilously close to the general warrants the Framers abhorred.[11] On the other side, cases like Rhine and Wells insisted that if the geofence is narrow enough, and the process sufficiently constrained, traditional standards of probable cause and particularity can be satisfied.[12]
For future investigations relying on LHD, the real legal fight won’t be about “reverse” warrants to a giant corporate database. It will return to more familiar ground: warrants to search particular devices and accounts, governed by the Supreme Court’s 2014 ruling in Riley v. California. In Riley, the Court was unequivocal: phones are “a digital record of nearly every aspect of [our] lives,” and police generally need a warrant to search them.[13] By moving LHD onto phones, Google has ensured that, going forward, access to a person’s historical location record will almost always require a device‑specific warrant.
Third, Google’s change paradoxically strengthens one of the most promising ways courts have tried to reconcile geofence warrants with privacy: a timing‑based framework that treats different stages of the process differently. Under that approach, the initial, anonymized query, Step One, can be governed by statute and the third‑party doctrine, as long as it is tightly bounded in time and space. The real Fourth Amendment “bite” came later, at the point of de‑anonymization, Step Three.
As long as Sensorvault existed, that framework suggested a simple rule: lower legal process for anonymized database queries; a full, probable‑cause warrant for de‑anonymizing specific devices and expanding the time window. Courts could demand a second, more particularized warrant before Google handed over names and account details.
Once Location History moves onto devices, that timing‑based rule and the technology itself converge. The act of de‑anonymization will now almost always happen through a traditional warrant to search that person’s phone or account. The world that critics feared, in which a single order to a tech giant could disgorge names for everyone near a protest or clinic, becomes harder to achieve as a matter of engineering.
It would be a mistake, though, to think that Google’s storage change renders the Supreme Court’s forthcoming decision in Chatrie irrelevant. First, there are the many past cases still being litigated, including Chatrie’s own. The Court will have to say something about whether, and on what terms, reverse‑location warrants to a centralized database were constitutional when they were used. Its answer will matter for defendants whose convictions rest on that evidence.
Second, Sensorvault may be going away, but the investigative impulse behind geofence warrants will not. If law enforcement cannot get a reverse‑location list from Google, they may turn to cellular carriers, to app developers, or to other large platforms that log user movements. Those systems may not look exactly like Sensorvault, but the legal questions will be the same: when is it acceptable to sweep in data about many people at once, when must the government obtain a warrant, and at what point does a “big data” search become a forbidden general warrant in digital form?
Finally, Google’s move is a vivid reminder that design choices by private companies can reroute constitutional law in practice. For several years, courts, scholars, and police departments struggled to draw lines around geofence warrants. Then one company re‑architected its storage, and the most aggressive version of the technique largely disappeared. That should not substitute for legal limits; there is no guarantee the next platform will make the same choice. It does, however, suggest a more modest role for the courts in this space: rather than trying to bless or ban an entire investigative method in the abstract, they might (as I argued) focus on the points of greatest intrusion—the moments when anonymous information becomes personal, names are attached, and movements over time are assembled into a narrative of a life.
The Supreme Court’s ruling in Chatrie will arrive in a world where the original Google geofence is already fading. Yet the core issue endures: how to adapt centuries‑old protections against unreasonable searches to a landscape where our movements are routinely recorded and where the line between helpful lead and intrusive dragnet can be crossed with a few keystrokes. Google’s decision to retreat from central storage does not answer that question. It merely ensures that, whatever the Court decides, the most important searches in the geofence story will still take place where they always have: on our own persons, in our own pockets, and subject to a warrant and the watchful eye of the United States Constitution.
[1] Ronald J. Rychlak, Geofence Warrants: The New Boundary, 93 Miss. L.J. 957 (2024).
[2] United States v. Smith, 110 F.4th 817, 838 (5th Cir. 2024).
[3] In re Search of Information Stored at Premises Controlled by Google LLC, 579 F. Supp. 3d 62 (D.D.C. 2021); United States v. Rhine, 652 F. Supp. 3d 38 (D.D.C. 2023); Wells v. State, 714 S.W.3d 614 (Tex. Crim. App.), reh’g denied, 721 S.W.3d 260 (Tex. Crim. App. 2025).
[4] Chatrie v. United States, 223 L.Ed.2d 553 (U.S. 2026).
[5] United States v. Chatrie, 107 F.4th 319 (4th Cir. 2024).
[6] Id. at 325.
[7] Id.
[8] United States v. Chatrie, 136 F.4th 100 (4th Cir. 2025).
[9] Chatrie v. United States, 223 L.Ed.2d 553 (U.S. 2026).
[10] Kylie Kirschner, Google will no longer hold onto people’s location data in Google Maps — meaning it can’t turn that info over to the police, Business Insider (Dec. 15, 2023), https://www.businessinsider.com/google-maps-location-data-history-stored-locally-2023-12.
[11] Smith, 110 F.4th at 838.
[12] Rhine, 652 F. Supp. 3d at 89; Wells, 714 S.W.3d at 626.
[13] Riley v. California, 573 US 373 (2014).
MEET THE AUTHOR
Ronald J. Rychlak is Distinguished University Professor and University of Mississippi School of Law’s Jamie L. Whitten Chair in Law and Government, where he has taught since 1987. A former Associate Dean for Academic Affairs and longtime Faculty Athletics Representative, he also serves as legal advisor to the Holy See’s Mission to the United Nations. Professor Rychlak is the author or editor of twelve books and more than 100 articles, with scholarship spanning criminal law, constitutional law, evidence, gaming law, and international law. His work has appeared in leading journals and national media outlets, and he is a frequent commentator on issues at the intersection of law, religion, and public policy.