Geofence Warrants – Part 5: Google’s 2023 Policy Shift
Google announced it would stop storing location data centrally, effectively ending its role as a geofence warrant target. A year later, investigators and civil liberties groups aren’t sure the fix is real—or whether it matters.
—
On December 8, 2023, Google announced a policy change that was, depending on your perspective, either a landmark privacy victory or a sophisticated deflection. The company would stop storing users’ location data in Sensorvault—its central database that had served as the primary target for more than 11,000 geofence warrants per year.[1] Instead, location data would be stored on users’ devices, encrypted, with Google no longer holding the centralized repository that made geofence warrants possible.
A current Google employee, not authorized to speak publicly, confirmed to Forbes that the move was explicitly designed to bring an end to geofence warrants using Google information.[2] Police Magazine reported that “along with the obvious privacy benefits of encrypting location data, Google made the move to explicitly bring an end to geofence warrants using Google information.”[3] The Federal Public Defender for the District of Kansas asked the obvious question: “Did Google End Geofencing Warrants?”[4]
The answer, a year and a half later, is: maybe. And maybe it doesn’t matter.
—
What Google Announced
Google’s announcement was straightforward in its framing but limited in its details. Location History data—the core of the Sensorvault database—would be stored on-device rather than in Google’s central servers. Users could still enable or disable Location History, but the data would reside on their phones, encrypted, rather than in a Google-maintained database.[5]
The immediate effect was clear: if Google doesn’t hold centralized location data, it cannot comply with a geofence warrant directing it to search that data. Step 1 of the geofence process—Google searching Sensorvault for all devices in a geographic area during a time window—requires a central database. Without one, the warrant returns nothing.
But the technical implementation raises questions that Google has not publicly addressed. If location data is encrypted on-device, who holds the encryption keys? Can Google decrypt the data remotely if compelled by court order? Is the data truly inaccessible to Google, or is it inaccessible only to routine law enforcement requests that don’t involve compelled decryption? The difference matters: a system where Google holds the keys is constitutionally different from one where it doesn’t, and the Fourth Amendment analysis changes depending on which architecture Google actually implemented.[5a]
The announcement was widely covered as a victory for privacy. The Forbes investigation in October 2024, however, raised questions about whether the policy change was as complete as Google suggested—and whether privacy victories won through corporate policy are victories at all.
—
The Forbes Investigation: Unanswered Questions
In October 2024, Forbes published “Google To Stop Giving Location Evidence To Law Enforcement,” an investigation that examined whether Google’s policy change had actually been implemented as described.[6] The investigation raised several concerns:
Is the data really decentralized? Google’s announcement said location data would be stored on-device. But Google Maps’ real-time traffic data—crowdsourced from users who enable location sharing—is generated by collecting aggregate location data from many devices simultaneously. If that data exists on Google’s servers in any form, it could potentially be accessed through a different type of warrant.[7] The Forbes investigation noted that even with on-device storage, the technical question of whether Google retains the capability to reconstruct centralized location data—through server-side logs, backup systems, or real-time traffic aggregation—remains unanswered.
What about Android devices? Android phones with Location History enabled continue to collect location data. Even if that data is not stored centrally in Sensorvault, it exists on individual devices. Law enforcement could, in theory, obtain warrants for individual devices—if they know whose device to target. But the whole premise of a geofence warrant is that the target is unknown. The policy change may have eliminated the efficient form of geofence surveillance without eliminating the possibility of location data access.[8] However, as the Colorado Law model framework analysis noted, law enforcement is already pivoting to alternative data sources—mobile carriers, app developers, and data brokers—that maintain centralized location records outside Google’s ecosystem.[8a]
What about third-party apps? Many apps use Google’s location services API, which collects location data as part of its functionality. This data may be stored on Google’s servers separate from the Location History database that was the target of geofence warrants. If law enforcement can access this data through a different legal process, the policy change may simply redirect geofence warrants to a different data source.[9]
What about Google Photos? Photos taken on Android devices often include embedded geolocation metadata. This data exists on Google’s servers when photos are backed up to Google Photos. A warrant targeting Google Photos metadata could achieve results similar to a geofence warrant, depending on how the data is stored and indexed.[10]
Google has not publicly disclosed the full architecture of its location data systems after the policy change. The company’s transparency reports, which previously disclosed geofence warrant volume, have not provided updated data on whether geofence warrant requests have declined since December 2023.[11]
—
The Federal Public Defender’s Assessment
The Federal Public Defender for the District of Kansas, writing shortly after Google’s announcement, identified a critical limitation of the policy change: it does not affect historical data.[12]
The geofence warrant that identified Okello Chatrie was served in 2019 and searched data from 2019. Even if Google’s policy change eliminates future geofence warrant vulnerability, location data collected before December 2023 may still exist in Google’s systems—and may still be accessible through warrants. The policy change is prospective, not retroactive.
This matters because geofence warrants are often used in cold cases—investigations of crimes that occurred months or years before the warrant is served. A 2024 geofence warrant could seek 2019 location data. If that data still exists in Sensorvault or a successor system, the policy change does not prevent its disclosure.[13]
The Federal Public Defender’s analysis also raised a question about Google’s data retention practices that has not been publicly answered: does Google retain backup copies of Sensorvault data for disaster recovery or regulatory compliance purposes? If so, those backups could be subject to geofence warrants even if the “live” database has been discontinued. Google has not disclosed its data retention schedule for historical Sensorvault records, and the company’s transparency reports do not address whether pre-2023 location data remains accessible through legal process.
The Federal Public Defender also noted that the policy change is just that—a policy. Google could reverse it at any time. There is no statutory or judicial mandate requiring Google to decentralize its location data. The constitutional question—whether geofence warrants violate the Fourth Amendment—remains regardless of whether Google’s current business practices make them more or less convenient to execute.
—
Does It Matter If the Fix Is Real?
Even if Google’s policy change is fully implemented and effective, it addresses only one vector of the geofence surveillance problem. Location data exists in multiple repositories, collected by multiple companies, accessible through multiple legal processes. Eliminating the largest target—Sensorvault—reduces the convenience of geofence surveillance without eliminating the capability.
More fundamentally, Google’s policy change highlights a structural problem with relying on corporate self-regulation to address constitutional concerns. The Fourth Amendment protects against unreasonable searches and seizures by the government. When the government serves a warrant on a private company, the constitutional question is whether the warrant is valid—not whether the company chooses to comply. If the Constitution permits geofence warrants, then Google’s refusal to comply is a business decision, not a constitutional protection. If the Constitution prohibits geofence warrants, then Google’s policy change is redundant—though practically welcome.
The Chatrie case itself illustrates the point. The geofence warrant that identified Chatrie was served on Google in 2019. The constitutional question—whether that warrant violated the Fourth Amendment—has nothing to do with Google’s 2023 policy change. The Supreme Court will answer that question based on constitutional law, not corporate policy.
But the policy change does have implications for the practical future of geofence surveillance. If Google no longer holds centralized location data, the primary target for geofence warrants disappears—at least until another company builds an equivalent database, or until law enforcement finds a way to access decentralized data through alternative legal processes.
—
Alternative Data Sources: The Next Targets
Even if Google’s decentralized storage eliminates Sensorvault as a geofence target, the underlying capability—searching a large database of location records without knowing who you’re looking for—exists in other forms:
Telecommunications carriers still hold CSLI data—the same data at issue in Carpenter. A geofence warrant served on AT&T or Verizon could achieve similar results using cell-tower records, though with less geographic precision than GPS-based Location History.[14a]
Connected vehicles increasingly transmit location data to manufacturers. General Motors’ OnStar system, Tesla’s Autopilot logs, and similar systems create location records that could be targeted through geofence warrants.[14b]
Smart city infrastructure—traffic cameras, license plate readers, transit systems—generates location data at municipal scale. These systems are government-owned, making constitutional challenges more complex because there is no corporate intermediary to resist the warrant.[14c]
The geofence warrant question, stripped of its Google-specific framing, is really a question about the constitutional limits of searching any large database of location records without individualized suspicion. Google’s policy change eliminates one target. The constitutional question applies to all of them.
—
The Limits of Corporate Self-Regulation
Google’s policy change illustrates a deeper problem with relying on corporate decisions to solve constitutional questions. The Fourth Amendment constrains the government, not private companies. When Google decides to stop storing location data centrally, it makes geofence warrants less convenient—but it doesn’t make them less constitutional. If the Fourth Amendment permits geofence warrants, then a future Google policy reversal (or a competitor’s decision to centralize location data) could make them convenient again overnight.
The Institute for Justice, in its March 2026 amicus brief, framed the problem precisely: constitutional rights should not depend on corporate data architecture.[14d] Whether the government can search a database of location records without individualized suspicion is a question of constitutional law, not corporate policy. If the answer is no, then geofence warrants are unconstitutional regardless of who holds the data. If the answer is yes, then no corporate policy change can provide lasting protection.
This is why the Supreme Court’s decision in Chatrie matters more than Google’s policy change. A constitutional ruling creates a binding precedent that applies to every company and every database. A corporate policy creates a temporary barrier that can be reversed, circumvented, or rendered obsolete by technological change.
—
Google’s Constitutional Evolution: From Compliance to Opposition
Google’s relationship with geofence warrants has undergone a remarkable transformation—one that itself illustrates the limits of corporate self-regulation and the need for constitutional resolution.
When geofence warrants first emerged, Google complied. The company built internal systems to process them efficiently, creating the three-step procedure that became standard. Google was, in effect, the enabler of the geofence warrant regime: without its centralized Sensorvault database and cooperative process, the warrants could not have functioned at the scale they reached—11,500+ in 2020 alone.
Over time, Google’s posture shifted from compliance to resistance. The company began moving to quash geofence warrants in individual cases, arguing they were overbroad. By the time it filed its Supreme Court amicus brief in Chatrie on March 2, 2026, Google had objected to over 3,000 geofence warrants on constitutional grounds.[14e]
Google’s amicus brief, filed in support of neither party, marks the fullest expression of this evolution. The company did not take a position on whether the specific warrant in Chatrie’s case satisfied the Fourth Amendment. But it argued forcefully that law enforcement was required to obtain a warrant to access Location History data—and that the third-party doctrine should not strip users of Fourth Amendment protections merely because they stored data with Google.[14f]
The brief’s core argument reframes Location History not as a business record held by Google, but as a user’s personal record: “Individuals’ documents and data stored electronically and securely in remote servers are the modern-day ‘papers[] and effects’ protected by the Fourth Amendment.”[14g] After Carpenter, Google argued, there can be “little question” that individuals have a reasonable expectation of privacy in their physical movements—and Location History was “often more precise and more comprehensive than the cell-site location information at issue in Carpenter.”[14h]
This framing is significant because it would protect Location History data even if the third-party doctrine applies—a question the Fourth Circuit answered yes and the Fifth Circuit answered no. If Location History is a user’s “papers” rather than Google’s business records, then the user retains Fourth Amendment rights regardless of whether sharing data with a third party eliminates those rights. The brief also emphasized that Location History was off by default and only collected if a user chose to enable it, distinguishing it from the passive data collection in Smith v. Maryland.[14i]
Google’s evolution—from building the infrastructure that made geofence warrants possible, to objecting to 3,000+ of them on constitutional grounds, to urging the Supreme Court to recognize Fourth Amendment protection for cloud-stored location data—traces the arc of a company grappling with the consequences of its own technology. But it also underscores the structural problem: Google’s position could change again. A different CEO, a different business model, a different regulatory environment could reverse the trajectory overnight. The Constitution does not change with corporate strategy. That is why Chatrie matters.
—
The Mootness Question
Google’s policy change raised a practical question for the Chatrie litigation: does it make the case moot? If Google no longer stores location data centrally, is there a live controversy about whether geofence warrants targeting Sensorvault are constitutional?
The answer is no, for three reasons. First, the Chatrie warrant was served in 2019 and searched historical data. Google’s 2023 policy change is prospective—it does not eliminate the historical data that was the subject of the 2019 warrant. The constitutional question about that warrant remains live regardless of Google’s current data architecture.
Second, even if Google has decentralized its data, the constitutional question extends beyond Google. Telecommunications carriers, app developers, connected vehicle manufacturers, and data brokers all hold location data that could be targeted through geofence warrants. A ruling that addresses only Google’s Sensorvault would be obsolete before the ink dries. The Supreme Court granted certiorari to address the constitutionality of geofence warrants as a category, not as applied to a single company’s database.
Third, the government argued in its certiorari opposition that the case presents a question of “exceptional importance” that recurs frequently. The Court agreed. The constitutional question does not depend on whether any particular company continues to hold centralized location data; it depends on whether the Fourth Amendment permits the process that geofence warrants use—searching a database without individualized suspicion, then generating suspects from the results.
The Institute for Justice, in its March 2026 amicus brief in Chatrie, framed the constitutional question precisely: “Multi-step warrant violates particularity requirement; generalized searches of innocent bystanders unconstitutional; analogizes to general warrants.”[14] The question is not whether Google makes geofence warrants convenient. The question is whether the Constitution permits them at all.
The Supreme Court’s answer—and its implications far beyond geofence warrants—is the subject of Part 6.
—
Notes
-
Google, announcement regarding Location History policy change (Dec. 8, 2023). See Federal Public Defender, District of Kansas, “Did Google End Geofencing Warrants?” (Dec. 21, 2023), https://ks.fd.org/blog/2023/12/8667-did-google-end-geofencing-warrants. ↩
-
Forbes, “Google To Stop Giving Location Evidence To Law Enforcement” (Oct. 8, 2024), https://www.forbes.com/sites/larsdaniel/2024/10/08/google-to-stop-sharing-location-data-with-law-enforcement/ (citing current Google employee confirmation). ↩
-
*Police Magazine*, “Google to No Longer Respond to Geofence Warrants,” https://www.policemag.com/news/google-to-no-longer-respond-to-geofence-warrants (quoting Google employee confirming policy designed to end geofence warrants). ↩
-
Federal Public Defender, District of Kansas, “Did Google End Geofencing Warrants?” (Dec. 21, 2023), https://ks.fd.org/blog/2023/12/8667-did-google-end-geofencing-warrants. ↩
-
See University of Minnesota LawSci Forum, “Caught in the Digital Dragnet: The Controversy Over Geofence Warrants and Privacy Rights” (Jan. 2025), https://mjlst.lib.umn.edu/2025/01/20/caught-in-the-digital-dragnet-the-controversy-over-geofence-warrants-and-privacy-rights/ (describing Google’s 2023 policy change: “location data would be stored on user devices” rather than centrally). ↩
-
Forbes, “Google To Stop Giving Location Evidence To Law Enforcement” (Oct. 8, 2024), https://www.forbes.com/sites/larsdaniel/2024/10/08/google-to-stop-sharing-location-data-with-law-enforcement/. ↩
-
Id. (raising questions about Google Maps real-time traffic data as a residual source of centralized location data). ↩
-
See Federal Public Defender, District of Kansas, “Did Google End Geofencing Warrants?” (Dec. 21, 2023), https://ks.fd.org/blog/2023/12/8667-did-google-end-geofencing-warrants (analyzing limitations of on-device storage as a barrier to geofence warrants). ↩
-
Id.; see also Forbes investigation (Oct. 2024) (discussing third-party app location data as potential alternative target). ↩
-
Id. (discussing Google Photos geotagged images as residual location data source). ↩
-
Google has not disclosed post-2023 geofence warrant volume data as of April 2026. ↩
-
Federal Public Defender, District of Kansas, “Did Google End Geofencing Warrants?” (Dec. 21, 2023), https://ks.fd.org/blog/2023/12/8667-did-google-end-geofencing-warrants (noting that the policy change does not affect historical location data). ↩
-
Id. ↩
-
Institute for Justice, Amicus Brief in *Chatrie v. United States* (Mar. 2, 2026), https://ij.org/amicus/chatrie-v-united-states/ (arguing that multi-step geofence warrant process violates particularity requirement and constitutes a general warrant). ↩
-
See generally Riana Pfefferkorn, “The Demise of Google’s Sensorvault and the Future of Geofence Warrants,” 52 *AIPLA Q.J.* 1 (2024) (analyzing the technical and constitutional implications of on-device encryption for geofence warrant access). ↩
-
Vivek Krishnamurthy, “A Model Framework for Regulating Geofence Warrants,” *Tech Policy Press* (June 9, 2025), https://www.techpolicy.press/a-model-framework-for-regulating-geofence-warrants/ (noting that law enforcement will “pivot to other data sources” as Google decentralizes location data). ↩
-
See *Carpenter v. United States*, 585 U.S. ___, 138 S. Ct. 2206 (2018) (holding that government acquisition of CSLI from wireless carriers constitutes a Fourth Amendment search). ↩
-
See generally Andrew J. Cichonski, “Connected Vehicles and the Fourth Amendment: Location Data in the Age of the Internet of Things,” 97 *Wash. U. L. Rev.* 549 (2020) (analyzing location data collection by connected vehicle systems). ↩
-
See generally Bennett Cyphers, “A Proliferation of Surveillance Technology Is Reshaping U.S. Policing,” *Electronic Frontier Foundation* (Apr. 2023), https://www.eff.org/deeplinks/2023/04/proliferation-surveillance-technology-reshaping-us-policing (documenting municipal surveillance systems including ALPR networks and smart city infrastructure). ↩
-
Institute for Justice, Amicus Brief in *Chatrie v. United States* (Mar. 2, 2026), https://ij.org/amicus/chatrie-v-united-states/ (arguing that constitutional protections cannot depend on corporate data architecture decisions). ↩
-
Brief for Amicus Curiae Google LLC in Support of Neither Party, *Chatrie v. United States*, No. 25-112 (U.S. Mar. 2, 2026), https://www.supremecourt.gov/DocketPDF/25/25-112/399674/20260302134430681_25-112%20Google%20Chatrie%20Amicus.final.pdf (stating that Google has “objected to over 3,000 geofence warrants on constitutional grounds”). ↩
-
Id. at 3–7 (arguing that “law enforcement was required to obtain a warrant to access [Location History] data” and that the third-party doctrine should not apply to user data stored by Google). ↩
-
Id. at 3 (“Individuals’ documents and data stored electronically and securely in remote servers are the modern-day ‘papers[] and effects’ protected by the Fourth Amendment.”). ↩
-
Id. at 3–4 (arguing that after *Carpenter*, “there can be little question that individuals have a reasonable expectation of privacy in the whole of their physical movements,” and that Location History was “often more precise and more comprehensive than the cell-site location information at issue in *Carpenter*”). ↩
-
Id. at 4 (noting that Location History was “off by default and was only collected if a user chose to enable it,” distinguishing it from the passive pen register collection in *Smith v. Maryland*, 442 U.S. 735 (1979)). ↩