Blog
Location-Based (Examples)
Location Data as Criminal Evidence
The phone places you at the scene before anyone identifies your face. In most of the cases that reach this firm involving location evidence, the defendant’s first encounter with the accusation is not a witness statement or a surveillance image but a data point: a coordinate, a timestamp, a cell tower record that the prosecution presents as a confession of presence. The evidence is assembled, interpreted, and offered as fact before the person it concerns has an opportunity to examine its accuracy or its origin.
What makes location evidence unusual is not its precision but the confidence it inspires in prosecutors who have never questioned it. A cell site location record is not a photograph of what occurred. It captures proximity, within a margin of error that varies depending on tower density and terrain. You do not contest a charge by ignoring the data that produced it. The distinction between presence and proximity is the one most juries are never invited to examine.
Historical Cell Site Location Information
Every cellular telephone, by virtue of its operation, maintains a continuous dialogue with the nearest cell towers. Each connection generates a time-stamped record. These records, known as cell site location information, accumulate without any affirmative act by the user beyond carrying the device. The Supreme Court in Carpenter v. United States recognized this quality when it observed that CSLI provides the government with something that approaches a comprehensive catalogue of a person’s movements, limited only by the retention policies of the wireless carrier.
Carpenter resolved one question and deferred several others. The Court held that obtaining seven or more days of historical CSLI constitutes a Fourth Amendment search requiring a warrant. The opinion was careful, almost conspicuously so, to confine its reach. It declined to address real-time CSLI, tower dumps, shorter periods of historical data, and the situation where the government seeks not the records of a particular suspect but the records of every person in a geographic area during a particular interval.
That last omission is the one that matters now.
The practical consequence of Carpenter was adaptation. The warrant requirement for historical CSLI became routine. Prosecutors continued to present cell tower evidence at trial. Juries continued to receive that evidence as something close to proof of location. The underlying fragility of CSLI (the margin of error, the difference between a tower’s serving sector and a device’s actual position, the effects of signal reflection in congested environments) remained unexamined in most courtrooms. The evidence enters the record uncontested, and the jury treats it as though it were a fingerprint.
In this firm’s practice, the accuracy of CSLI is challenged in fewer than one in five cases where it is introduced. The reasons are not complicated. Challenging CSLI requires a technical expert in cell site analysis, and such experts are expensive and difficult to retain. Defense attorneys handling appointed cases seldom have the resources to engage one. The government’s evidence goes unchallenged, and it should not.
It is not.
CSLI accuracy depends on tower density, terrain, atmospheric conditions, network load, and the specific technology the carrier employs to log connections. In a case last year in the Eastern District, a tower serving a radius of roughly two miles was offered to the jury as evidence placing the defendant at a specific address. The government’s analyst testified that the data was “consistent with” the defendant’s presence at that location, a phrase that does a considerable amount of work in these proceedings, and one that permitted the jury to treat approximation as certainty. The signal evidence was also consistent with the defendant’s presence at a grocery store, a public park, and something like nine hundred residences within the same serving area.
The Warrant That Named No One
A geofence warrant reverses the traditional investigative sequence. Conventional warrants identify a suspect and seek evidence concerning that person. A geofence warrant identifies a place and a time and seeks the identity of every person whose device was present. The warrant does not name a suspect. The suspect is what the warrant is designed to produce.
The process, at least as it operated through Google’s Sensorvault database, followed a three-step protocol. The government defined a geographic boundary and a temporal window. Google returned an anonymized list of devices present within those parameters. The government then narrowed that list and requested identifying information for selected accounts. The procedure was straightforward. The constitutional questions it raised were not.
From the first geofence request in 2016, the volume of these warrants increased by orders of magnitude in subsequent years. Google received thousands of such requests annually, covering offenses from armed robbery to property damage to stolen vehicles. The warrants swept in every person who happened to carry a device through the designated area, a population that in most cases had no connection to the crime under investigation.
Whether a person who walks past a bank at eleven in the morning has forfeited a constitutional interest in keeping that movement private is a question the Court had not, until this term, agreed to consider.
The Circuit Split and Chatrie v. United States
Okello Chatrie was charged with robbing a bank in Richmond, Virginia. The government identified him through a geofence warrant served on Google, which returned location data for every device present near the bank during the hour surrounding the robbery. The warrant did not name Chatrie. It did not identify his device. It identified a geography and asked Google to search through its records, hundreds of millions of accounts, to determine who was there.
The district court in the Eastern District of Virginia found the warrant unconstitutional. It held that the government lacked particularized probable cause as to every Google user captured in the geofence. But the court declined to suppress the evidence, invoking the good faith exception: the officers had acted in reasonable reliance on a judicial authorization that no appellate court had yet invalidated.
The Fourth Circuit’s handling of the case produced something close to jurisprudential vertigo. A three-judge panel held that Chatrie had no reasonable expectation of privacy in two hours of location data voluntarily shared with Google, sidestepping the constitutional question. On rehearing en banc, the full court vacated that panel opinion and issued, in April 2025, a single-sentence per curiam affirmance of the district court. The opinions that accompanied it ran to over a hundred pages, with nine separate concurrences and dissents. The en banc court split seven to seven on whether a search had occurred at all.
The Fifth Circuit, confronting a similar warrant in United States v. Smith, reached the opposite conclusion. It held that geofence warrants are the modern equivalent of general warrants and unconstitutional under the Fourth Amendment. Yet it too applied the good faith exception. Two circuits, two constitutional frameworks, and the same result: the evidence comes in.
I drafted this passage in late March, with oral argument a month away, which may account for its anticipatory quality. The Supreme Court granted certiorari in Chatrie in January 2026, and argument is scheduled for April 27. The government’s position (which, it should be noted, rests on a reading of the third-party doctrine that the Court in Carpenter went to considerable lengths to limit, a reading that treats voluntary opt-in to a mapping service as equivalent to leaving physical evidence at a crime scene, and that conflates a database of hundreds of millions of users’ movements with the kind of discrete business record the doctrine was designed to address) is that Chatrie diminished his own expectation of privacy. The defense, supported by civil liberties groups, technology firms, and press freedom organizations, argues that geofence warrants fail the Fourth Amendment’s particularity requirement because they cannot, by design, describe the person to be searched.
I am less certain than the commentary suggests about which way this Court will rule. The narrow path is available: decide the case on its facts, leave the broader question open, and let the doctrine develop in the lower courts for another decade. The Court could rule narrowly, on the specific facts of Chatrie’s case, and leave the broader question for another term, which is what the Court did in Carpenter. The unanswered questions from that decision are what produced the circuit split that Chatrie is supposed to resolve.
Retention Policies and Storage Changes
In late 2023, Google announced that it would begin storing Location History data on users’ devices rather than in its centralized Sensorvault database. The default retention period was reduced to three months. Backed-up data would receive encryption, with keys held only by the user. The changes took effect in late 2024.
The practical effect is that Google can no longer comply with most geofence warrants because it no longer possesses the centralized data those warrants require. Law enforcement agencies have noted the change. Whether they will redirect requests to other providers, or develop alternative methods for obtaining bulk location data, remains unclear. Cell carriers still collect tower-level location information through ordinary network operations. That data is less precise than what Sensorvault provided, but it is not nothing.
The legal principle governing whether the government may compel a technology company to search its users’ location records remains the question that the storage change, for all its practical significance, did not touch.
Challenging Location Evidence
Suppression of location evidence begins with the warrant. If the warrant was issued without probable cause, or if it fails to describe with particularity the place to be searched and the persons or things to be seized, the evidence it produced is open to challenge. In geofence cases, the particularity argument is the strongest one available, because the warrant by design does not and cannot identify the person whose data is sought.
The good faith exception functions, in this area, the way a smoke detector functions in a building that has already been condemned: technically present, operationally something less than reassuring. Courts in both Chatrie and Smith found the warrants unconstitutional and declined to suppress the evidence. Until the Supreme Court establishes a clear rule, good faith will continue to serve as a safety net for prosecution.
But there is a second path, and it is the one this firm takes in every case involving cell site or geofence evidence. The legal challenge to the warrant is necessary. It is not sufficient. The technical challenge to the evidence itself is where suppression motions succeed or fail in practice.
There are steps that defense counsel should take early:
- Obtain the complete warrant application, including the affidavit and any attachments describing the geofence parameters or the cell towers at issue.
- Request the provider’s raw data: tower identifiers, signal strength measurements, timing advance values, and the methodology the government’s analyst used to estimate location.
- Retain or consult a cell site analysis expert before the suppression deadline, not after.
That third step is the one most frequently omitted. A motion to suppress that challenges the legal basis for a warrant without contesting the reliability of the underlying data leaves the prosecution’s strongest evidence intact. In our practice, we file the suppression motion and the motion to exclude the government’s technical testimony as companion filings, because a judge who denies suppression on good faith grounds will sometimes still limit what the analyst may say to the jury if the methodology is shown to be unreliable. The clerks in this district, if we are being precise, tend to schedule the technical hearing first, which gives us an opportunity to establish the data’s limitations before the constitutional argument is heard.
The difference between “this evidence should not have been obtained” and “this evidence does not demonstrate what the government claims” is the difference between a constitutional argument and a practical one. Most cases turn on the practical argument, even if the constitutional one occupies more space in the briefs.
Every person carrying a device that records where it has been possesses a stake in where the law draws this boundary, between the government’s interest in solving crimes and the individual’s interest in moving through a city without generating a searchable archive of every step. That boundary is, at the moment, being drawn by nine justices who will hear argument next month on a case that began with a bank robbery in Richmond and has arrived at a question about what kind of surveillance a free society is prepared to authorize.
A consultation with this firm begins that conversation, and it costs nothing.