Five years ago, I wrote about geofence warrants. I said then that “there are zero cases on Westlaw . . . [and] virtually no secondary source material about these warrants.” Times have changed. Now we have cases, including one from the Fourth Circuit, and lots of secondary source material. This post explains the state of the law on geofence warrants.
The basics of geofence warrants. Geofence warrants normally are used when police know that a crime occurred at a specific time and place, but do not yet have a suspect. These warrants require a technology company – almost always Google – to search its database of user location information and to produce information about users who were near the place in question around the time of the crime. Google receives tens of thousands of these warrants each year, because about one third of its registered users, meaning over 500 million people, have enabled a feature called Location History that more or less continuously shares the location of the user’s phone with Google. Google has developed a multi-step protocol for responding to these warrants. It initially shares anonymized data, then after law enforcement selects the accounts of greatest interest, shares users’ names and other identifying information.
What’s the controversy? Law enforcement views these warrants as vital tools for developing leads in cases where traditional investigative techniques have failed, and prosecutors often argue that any privacy intrusion is slight given the short time frame and limited geographic area at issue. Critics view these warrants as general warrants that authorize an overbroad search of mountains of data belonging to millions of users without particularized probable cause.
Recent cases. Two federal courts of appeals have decided cases about geofence warrants in the last few months.
The first is United States v. Chatrie, 107 F.4th 319 (4th Cir. 2024). Phil Dixon summarized Chatrie here, but in a nutshell, The defendant enabled the Location History feature of his Google account, then carried his phone with him when he robbed a credit union in Virginia. Investigators obtained a geofence warrant requiring Google to disclose information about devices present within 150 meters of the credit union between 30 minutes before the crime and 30 minutes after. Pursuant to Google’s standard process, the officer obtained progressively more detailed information (including information outside the initial temporal window and beyond the initial geofence) about progressively fewer devices, eventually obtaining the names and subscriber information of three account holders, including the defendant. This led to further investigation of the defendant, and eventually to federal criminal charges against him. He moved to suppress, and the district court denied the motion based on the good faith exception to the exclusionary rule. The defendant pled guilty and appealed. The Fourth Circuit affirmed on a different rationale, holding that the review and disclosure of Google’s records wasn’t a search at all under the third-party doctrine and so didn’t require a warrant. The majority stated that the defendant “did not have a reasonable expectation of privacy in two hours’ worth of Location History data voluntarily exposed to Google.” It distinguished Carpenter v. United States, 585 U.S. 296 (2018), the long-term historical cell site location information case, on the theory that the data at issue here was short-term (hours, not days) and much more voluntarily shared with Google (because users must opt in to providing Location History data). The court highlighted that only about a third of users share their location with Google and noted that enabling such tracking isn’t necessary to modern life in the way that Carpenter said having a cell phone is.
Judge Wynn wrote a long dissent arguing that Carpenter was revolutionary and requires a court considering whether an officer has conducted a search by obtaining digital information to assess the data’s “comprehensiveness, its retrospective capabilities that allowed for historical tracking, the intimacy of the information it reveals, and its ease of access . . . for police.” In Judge Wynn’s view, Location History is comprehensive, precise, and revealing, so obtaining it – even for a limited time period – is a search. And in footnote 12 of his dissent, he concludes that the warrant here did not support such a search because it lacked probable cause and particularity: “The government’s proposed justification—that the robber used a cell phone and a cell phone could have Google Location History turned on—is extremely broad.”
The second case is United States v. Smith, 110 F.4th 817 (5th Cir. 2024). In that case, the defendant robbed a driver working for the US Postal Service. Investigators obtained a geofence warrant for Google’s location data, hoping to identify individuals in the area at the time of the robbery. Google provided information indicating that the defendant had been in the area, and subsequent investigation led to criminal charges against him. He moved to suppress, the district court denied the motion, he was convicted at trial, and he appealed. The Fifth Circuit began its analysis by concluding that a geofence warrant entails a search. It noted that Location History is accurate and potentially intrusive, and while it is true that users must enable Location History, “electronic opt-in processes are hardly informed and, in many instances, may not even be voluntary.” Then the court determined that the warrant was inadequate to support the search because it was a general warrant that lacked particularity. The court emphasized that while the result of the search may be limited to a few users, the search itself includes all the data held by Google in connection with 592 million subscribers. However, the majority applied the good faith exception because the officers were acting cautiously and reasonably while attempting to use new investigative techniques.
These are major cases. Smith, in particular, would have major implications beyond geofencing because its rationale would apply to many cases in which investigators seek to query large databases, such as when investigators ask Google to identify users who searched for certain terms connected with a crime. Furthermore, Smith’s conclusion implies not just that the warrant in that case lacked probable cause and particularity, but that such warrants inherently lack probable cause and particularity and so – in essence – may never be issued.
Neither of these cases has fully concluded. The Fourth Circuit voted Friday to rehear Chatrie en banc, and a petition for rehearing en banc is pending in the Fifth Circuit concerning Smith.
Looking ahead. Since Chatrie and Smith arose, Google has changed how it handles location data. The Electronic Frontier Foundation explains here that “going forward, this data will be stored, by default, on a user’s device, instead of with Google in the cloud”; that it “will be set by default to delete after three months [rather than the current 18 months]”; and that “if users choose to back up their data to the cloud,” Google will encrypt it so that even Google can’t read it. Whether Google made these changes in order to protect users’ privacy or to reduce the burden of complying with geofence warrants, only Google knows for sure.
These changes may mean that Google receives fewer geofence warrants, and they reduce the likelihood that the Supreme Court will review this issue even if Chatrie and Smith end up as a circuit split. However, other technology companies are still storing users’ location information, and so-called tower dumps raise similar legal issues, so this is not a legal dead end.
It is also worth noting that by pushing Location History data onto users’ devices, Google may be making it easier for law enforcement to obtain search warrants for those devices. After all, if nothing else, a suspect’s phone is now more likely to contain location information linking the suspect to the scene of the crime.
Further reading. Readers interested in digging deeper may be interested in this law review note opposing geofence warrants and this blog post taking a more charitable view.