Case Study 18.2: Geofence Warrants and the Innocent Bystander Problem

Background

On a November afternoon in Phoenix, Arizona, a bank was robbed. Investigators quickly established the robbery's timing and obtained surveillance footage showing the robber's car — a dark-colored sedan, partial plate visible. They had a crime, a time, and a place. What they needed was a suspect.

The investigator in charge filed a geofence warrant with Google, requesting data on all Android devices that had been within a two-block radius of the bank between 1:00 PM and 2:30 PM. Google, which maintains location history data for users who have not disabled the feature through its Google Maps Timeline function, responded with data on 89 devices.

Among those 89 devices was the robber's phone. Also in the dataset: a food delivery driver who had made three deliveries in the area; a real estate agent who had shown a nearby property; two office workers whose employer was located within the geofenced area; a postal carrier on their regular route; a visiting nurse making a home health call; and 81 other people whose presence in the area at that time had nothing to do with the robbery.

The investigator requested additional location detail from Google on all 89 devices. Google provided a second tranche of data. Through a process of elimination — cross-referencing device movement patterns, filtering out obvious regular traffic (the office workers' devices showed the same workplace location daily) — the investigators narrowed the list to a handful of devices, including the robber's.

This is a geofence warrant working as designed: a broad data collection followed by narrowing to identify a suspect from an initial population of innocents. The question this case study examines is what happened to the 88 people who were not the robber — and what it means for a society to build a criminal justice infrastructure on the routine collection of data about innocent people.

What Happened to the 88

The 88 people whose data was collected had no notice that they had become subjects of a criminal investigation. Google, in compliance with the warrant, had provided their location data to law enforcement. Their data had been reviewed by investigators. Their daily patterns — where they worked, where they went after the robbery, what neighborhoods they lived in — had been assessed by the state.

None of them were suspects. None had committed any crime. Several had been at the relevant location for reasons that, if known to investigators or prosecutors, might themselves generate unwanted scrutiny: one person had been visiting an addiction treatment clinic in the area; another had been at a domestic dispute two blocks away that the responding officers did not know to connect to the geofence investigation.

None of them were notified that their data had been collected. None were told that they had been reviewed and excluded as suspects. None received the notification that criminal procedure typically requires when a person becomes the subject of investigation — because no single one of them, formally speaking, was a "suspect" at the time of the warrant.

The geofence warrant's architecture is designed precisely to permit this: mass data collection followed by analysis followed by narrowing. The broad initial collection is never classified as a "search" that triggers Fourth Amendment protections for the individuals swept up in it, because the warrant encompasses all of them equally — none is individually targeted until after the collection.

The Constitutional Question

In 2021, the Fifth Circuit Court of Appeals ruled in United States v. Chatrie that geofence warrants do not categorically violate the Fourth Amendment, applying a complex analysis that found the initial disclosure of data to be voluntary (because users had opted into Google's location history) and that applying the fourth amendment's reasonableness standard only when investigators obtained the second, more detailed tranche. The ruling was criticized by privacy scholars as analytically inconsistent — the "voluntary" disclosure relied on the same problematic interpretation of third-party doctrine that Carpenter had complicated.

In 2022, the Fourth Circuit ruled in United States v. Chatrie (the same defendant; the case was reheard after the Fifth Circuit's ruling created a circuit split) that the geofence warrant was unconstitutional. The circuit split meant the issue was likely to reach the Supreme Court.

The constitutional debate turns on several questions:

What is a "search"? The Fourth Amendment requires a warrant for a "search." If the government instructs a private company to review all location data for all devices in an area, is that a search — or is it merely a request to a third party, which the third-party doctrine exempts from warrant requirements?

When does the search occur? Google argues that it, not the government, reviews the initial dataset — that the government's search only begins when Google hands over the narrowed results. Critics argue that the government is using Google as a tool to perform a search the government itself would need a warrant to conduct directly.

What is the unit of analysis? Fourth Amendment analysis traditionally focuses on individual people or specific property. Geofence warrants operate at the population level — they are not searches of any particular person but of everyone in an area. The constitutional framework was not designed for this scale.

The Zachary McCoy Case

In January 2020, Zachary McCoy, a 30-year-old restaurant worker in Gainesville, Florida, received a notice from Google informing him that it had received a geofence warrant from local police investigating a residential burglary. McCoy had been in the area on his regular bicycle route, tracked by a fitness app. He had been to the street in question several times in the period covered by the warrant — because it was on his exercise route.

McCoy hired a lawyer, who was able to convince police that McCoy was a regular cyclist, not a burglar. The process cost McCoy several hundred dollars in legal fees, weeks of anxiety, and the specific experience of being a subject of a criminal investigation for being on a public street that happened to be the site of a crime he knew nothing about.

McCoy's case received significant press coverage because he was willing to speak publicly. But similar cases — innocent people swept into geofence warrant investigations and subsequently cleared — are structurally inevitable given how geofence warrants work. The warrant's design guarantees that the vast majority of devices captured will belong to innocent people. The investigative process of narrowing the dataset is the process of collecting and reviewing data about those innocent people.

What the Warrants Are Used For

The New York Times' 2020 reporting on geofence warrants found that the warrants had been used for crimes ranging from murder (where their use seems clearly proportionate to the investigative intrusion) to relatively minor property crimes (where the mass collection of innocent people's data seems disproportionate to the investigative benefit).

The same reporting found that geofence warrants had been requested in connection with political events — including investigations related to the January 6, 2021 attack on the U.S. Capitol, where investigators sought location data from devices present in the Capitol and surrounding areas. These warrants raised concerns not just about the individual investigation but about the chilling effect on political activity: if attendance at a politically charged location could sweep an innocent person into a criminal investigation, people might choose not to attend political events for self-protective reasons.

Reform Proposals

Several approaches to reforming geofence warrant practice have been proposed:

Limiting geographic scope: Requiring that the geofenced area be narrowly tailored to the specific crime scene, with judicial supervision of whether the proposed area is proportionate to investigative need.

Limiting temporal scope: Requiring that the time window be as narrow as possible given the investigative facts, rather than covering broad time windows that capture routine traffic.

Requiring narrow specificity before disclosure: Structuring the warrant process so that investigators must provide specific additional evidence before Google or other providers disclose identifying information, rather than receiving identifiable data in the initial response.

Notice requirements: Requiring notification to individuals whose data was collected and reviewed, even if they were excluded as suspects, so they can seek redress if they believe their data was misused.

Sunset and deletion requirements: Requiring that data collected through geofence warrants be deleted after a specified period if no criminal case results.

Proportionality requirement: Limiting geofence warrants to investigations of serious crimes, requiring courts to explicitly balance investigative benefit against the scope of innocent-person data collection.

Analysis

The Structural Transformation of Investigative Logic

Traditional criminal investigation begins with a suspect and gathers evidence. Geofence warrants reverse this: they begin with mass data collection and derive suspects from that collection. This reversal has been called "reverse warrants" by critics and represents a fundamental transformation in the logic of criminal investigation.

The reversal matters for several reasons. Individualized suspicion — the traditional requirement that investigation focus on specific people for specific reasons — serves as a check on government power: it prevents the state from conducting dragnet surveillance of large populations as a routine investigative tool. When investigation can begin without any individual suspicion, this check disappears.

The chilling effect that results — on presence at political events, medical facilities, religious gatherings — is not merely theoretical. It is the predictable response of rational individuals to a known risk of being swept into criminal investigations based on location.

The Commercial Infrastructure Problem

Geofence warrants are possible only because commercial data infrastructure — primarily Google's location history — retains enormous quantities of precise location data for large numbers of people. If Google did not retain location history data, geofence warrants would not exist as an investigative tool.

This raises a structural question about commercial data retention that individual warrant disputes do not address: should the commercial practice of retaining precise, comprehensive location histories for millions of users be permitted, given that it creates an infrastructure for mass government surveillance? The question is not whether geofence warrants are legal or constitutional; it is whether the underlying data infrastructure that makes them possible should exist.

Some scholars argue that the appropriate regulatory response to geofence warrants is not to require judicial approval for each warrant but to prohibit the commercial data retention that makes the warrants possible — implementing mandatory data minimization requirements for commercial location data that would prevent the accumulation of multi-month location histories for millions of users.

Discussion Questions

1. The Fifth Circuit found that Google's location history collection was "voluntary" because users had opted into the feature. Apply the "consent as fiction" framework from Chapters 16 and 18 to this legal finding. Is opting into Google Maps' Timeline feature a meaningful consent to eventual government access through a geofence warrant?

2. Zachary McCoy spent several hundred dollars and weeks of anxiety navigating an investigation in which he was never formally a suspect. What remedies should be available to innocent people whose data is collected in geofence warrant investigations?

3. The chapter notes that geofence warrants have been used for both serious crimes (murder) and minor property crimes. Should the scope of permissible geofence warrant use depend on the severity of the crime being investigated? Design a proportionality standard.

4. Critics argue that the appropriate response to geofence warrant concerns is commercial data minimization — prohibiting the retention of data that makes geofence warrants possible. Evaluate the tradeoffs: what would be lost if Google could not retain location history data? What would be gained?

5. The circuit split on geofence warrant constitutionality means the Supreme Court will likely decide the question. What constitutional analysis should the Court apply, and what outcome would best protect both legitimate law enforcement interests and the Fourth Amendment values at stake?