The Supreme Court heard oral arguments Monday in Chatrie v. United States, the first time the justices have taken a geofence warrant case directly. A decision is expected before the term ends.
The case is a strange one to land at the Supreme Court in 2026, because the technique it asks about no longer functions the way it did when the warrant in question was issued. Google ended its ability to respond to geofence warrants more than a year ago. The pipeline that produced more than 20,000 of these warrants between 2018 and 2020 is closed. The Court is ruling on a tool that, as a practical matter, is already gone.
Here is what the tool was, how it worked, and why it stopped working.
What A Geofence Warrant Is
A traditional search warrant names a suspect or a place. The government has probable cause that a person committed a crime, and the warrant authorizes a search of that person’s home, car or accounts.
A geofence warrant works the other direction. The government has no suspect. It has a location and a time window. It asks Google to identify every device whose Location History placed it inside that area during that window. Google runs the query and returns anonymized device IDs. The government narrows the list and asks Google to de-anonymize a handful.
The data is generated by Google Location History, the opt-in feature most people associate with Google Maps. Pre-cutover, that data could place a device within meters of its actual location at frequent intervals. That is closer to a GPS tracker than the cell tower data the Supreme Court ruled on in Carpenter v. United States in 2018.
How Often It Was Used
Geofence warrants moved from rare to routine fast. Google disclosed for the first time in 2021 that it had received 982 geofence warrants in 2018, 8,396 in 2019, and 11,554 in 2020. The 2020 total alone averaged more than 30 warrants per day.
The named cases span every type of investigation. Investigators used geofence warrants in postal robberies, in arson cases, homicides and in missing person searches. The largest single deployment was January 6. The FBI’s geofence warrant for a four-acre area around the Capitol on the afternoon of the riot returned 5,723 devices in Google’s initial sweep. Geofence data ran through many of the prosecutions that followed.
The technique was attractive precisely because it inverted the warrant process. A traditional warrant requires a suspect. A geofence warrant produces one.
How Google Quietly Ended The Geofence Era
In December 2023, Google announced that Location History would be saved locally on users’ devices instead of in their Google accounts, with the new feature renamed Timeline inside Google Maps. The default retention period dropped from 18 months to three. The rollout proceeded gradually through 2024 and was fully in place by December 1, 2024. I wrote about this earlier as Google was finalizing it, and the operative point then is the same now: once the data sits on the device, Google has no database to search.
The mechanics are simple. A geofence warrant orders Google to search its database for every device whose Location History placed it inside a defined area during a defined time. If Google does not hold that data, there is nothing to search. The Electronic Frontier Foundation, which has tracked geofence litigation across multiple circuits, made the same point: once Google moved the data on-device, the company could no longer respond to geofence warrants.
Pre-cutover data is a different question. Anything Google held in users’ accounts before the migration was already in litigation records or had been produced in response to earlier warrants. Chatrie’s case turns on data Google delivered in 2019.
The Facts In Chatrie
In May 2019, a man robbed the Call Federal Credit Union in Midlothian, Virginia at gunpoint and left with $195,000. The investigation stalled. The lead detective applied for a geofence warrant directing Google to produce Location History records for every device inside a 150-meter radius around the credit union for an hour bracketing the robbery. The 150-meter radius produces a 300-meter diameter, longer than three football fields. The area included a church, a hotel, a restaurant and a senior center.
Google’s response proceeded in three stages. Step one returned anonymized data on every device inside the geofence. Step two narrowed the list. Step three de-anonymized three devices and returned subscriber information. One was Okello Chatrie. He entered a conditional guilty plea preserving his right to challenge the warrant and was sentenced to roughly 12 years.
Two federal courts have already split on whether the warrant was constitutional. The full Fourth Circuit deadlocked 7-7 on whether a Fourth Amendment search had even happened, produced nine separate opinions across 126 pages, and affirmed Chatrie’s conviction in a single sentence. A different appeals court, the Fifth Circuit, looked at the same kind of warrant in United States v. Smith and ruled that geofence warrants violate the Fourth Amendment categorically. That disagreement is what the Supreme Court took up in January. The cert grant is narrow. The justices are evaluating this one warrant on this one set of facts, not whether geofence warrants are categorically banned.
What Happens Next
A ruling is expected before the term ends. Whatever the justices decide, it will not bring the geofence warrant back. Google does not hold the data anymore. The pipeline does not exist.
What the ruling will do is decide whether the warrants Google answered between 2018 and 2024 were constitutional in the first place, which still matters for any case that relied on geofence data and is still on appeal. Chatrie is the last major case of its kind from the Google geofence era. The decision in it will be the only Supreme Court word on the technique.
