Should the police be able to identify everyone who was in a busy metropolitan area, just because a crime occurred there? In two amicus briefs just filed in appellate courts, we argue that’s a clearly unconstitutional search.[1]

The two cases are People v. Meza, in the California Court of Appeal, and United States v. Chatrie, in the federal Fourth Circuit Court of Appeals. In each case, the defendant is challenging the police use of a surveillance tool we’ve written about before called a “geofence warrant.” In both cases, the lower courts denied motions to suppress. In Chatrie, however, the district court issued a lengthy opinion holding the geofence warrant was unconstitutional before ruling that police relied on the warrant in “good faith” and therefore the evidence from their search was admissible.

Unlike traditional warrants for electronic records, a geofence warrant doesn’t start with a particular suspect or even a device or account; instead police request data on every device in a given geographic area during a designated time period, regardless of whether the device owner has any connection to the crime under investigation. Google has said that for each warrant, it must search its entire database of users’ location history information—data on hundreds of millions of users.

The data Google provides to police in response to a geofence warrant has the potential to be very precise—much more precise than cell site location information, for example. It allows Google to determine where a user was at a given date and time, sometimes to within twenty meters or less. Google can even determine a user’s elevation and establish what floor of a building that user may have been on. As the lower court noted in Chatrie last summer, Google’s database “appears to be the most sweeping, granular, and comprehensive tool—to a significant degree—when it comes to collecting and storing location data.” At the same time, however, Google does not guarantee accuracy. Google’s goal is to accurately infer a user’s location within a certain radius a bare 68% of the time. This creates the possibility of both false positives and false negatives—people could be implicated for a crime when they were nowhere near the scene, or the actual perpetrator might not be included at all in the data Google provides to police.

The warrants in both the Meza and Chatrie cases encompassed large geographic areas and time periods. In Meza, the police asked for all devices in six discrete, heavily populated areas of Los Angeles during time periods where people were likely to be in sensitive places, like their homes at church or a medical center, or driving along one of the many busy streets included within the geofenced areas. In total, police requested data for a geographic area equivalent to about 24 football fields or five to six city blocks during five morning commute hours. Similarly, in Chatrie, the geographic area was about 17.5 acres (about 3 and a half times the footprint of a New York city block) and included a church, a chain restaurant, a hotel, several apartments and residences, a senior living facility, a self-storage business, and two busy streets.

In our briefs, we argue these warrants are unconstitutional “general warrants” because they don’t require police to show probable cause to believe any one device was somehow linked to the crime under investigation. Instead, they target everyone in the area and then provide police with unlimited discretion to determine who to investigate further. In Meza, we also argue the practice violates CalECPA, California’s landmark electronic communications privacy law.

Chatrie and Meza are the first cases challenging geofence warrants to make it to the appellate level. However, they appear to just be the tip of the iceberg. The number of police requests for geofence warrants has increased dramatically since their first reported use in 2016. According to Google, geofence requests now constitute more than a quarter of the total number of all warrants it receives, and 20% of those come just from law enforcement agencies in California.

There is real reason to be concerned about these overbroad searches. They have, in the past, caused innocent people to be suspected of crimes they didn’t commit. And geofence warrants can and have been used in ways that impact fundamental rights, including free speech and freedom of association. For example, during the protests following the police shooting of Jacob Blake, the ATF used at least 12 geofence warrants to collect people’s location data during protests in Kenosha, Wisconsin, one of which encompassed a third of a major public park for a two-hour window. Police also used a geofence warrant in Minneapolis around the time of the protests following the police killing of George Floyd. And geofence warrants may be used in the near future to target people for reproductive health choices and outcomes. Google has been sufficiently concerned about this possibility to pledge to delete location information shortly after someone visits an abortion clinic, though critics have argued this would be insufficient to protect people.

The Chatrie and Meza cases will both likely be argued sometime later this year. The majority of courts to address geofence warrants in publicly available opinions have raised constitutional concerns, refusing to issue the warrant or suppressing the evidence. We hope these two appellate courts will do the same.

[1] EFF was represented on the Chatrie brief by the NYU Technology Law & Policy Clinic, and the excellent brief was drafted by law students Talya Nevins and Yanan Wang.