In June 2025 federal investigators in Mississippi asked four carriers for every phone that connected to the towers near six crime scenes, 10 to 30 minutes of records at each, over a two-day span in 2024. They only wanted the phones that showed up at two or more of the towers. The magistrate judge said no. On August 5, 2026, U.S. District Judge Carlton Reeves agreed and went further, holding that no tower dump warrant can satisfy the Fourth Amendment.
Tower dumps have been a standard investigative tool for more than a decade, and the reason is the same thing that worries the courts: one request returns thousands of records, almost all of them belonging to people who were simply nearby. The investigative value is in the cross-reference. The constitutional problem is in the haystack. Both sides of that argument come down to how the records are handled after they arrive, which is a data problem before it is a legal one.
What a tower dump returns
A tower dump is not a record of one phone. It is a record of one tower, or one sector of it, for a window of time. The carrier returns every device that connected during that window, typically with the phone number and device identifiers, a timestamp, and the cell and sector that served the connection. The same carrier formats from the three carriers post apply: times in UTC from some carriers and local time from others, tower coordinates on the row or in a separate key, and sector information that has to be kept with the tower record. Those problems do not go away because the return is larger. They multiply.
How large? In the Mississippi applications the court estimated that towers near major highways would return thousands, if not tens of thousands, of records per location. Six locations means six of those lists. The phone the investigators are looking for, if it exists, is a handful of rows spread across them.
Where the courts are in 2026
The law here is moving, and anyone who handles these records should know the map.
- Carpenter v. United States (2018) held that historical cell site location information is protected by the Fourth Amendment, and expressly left tower dumps for another day.
- United States v. James (8th Cir. 2021) upheld a tower dump warrant, and other courts have done the same application by application.
- United States v. Smith (5th Cir. 2024) held that geofence warrants, at least as described in that case, are unconstitutional general warrants, because the search itself has to be particular, not just what investigators keep afterward.
- State v. Bryant (N.J. App. Div. 2025) held that tower dumps are searches that require a particularized warrant supported by probable cause. The New Jersey Supreme Court has agreed to hear the question.
- Chatrie v. United States (U.S. June 2026) held that geofence warrants are searches regardless of how narrow the time window is.
- In re Four Applications for Search Warrants (S.D. Miss. Aug. 5, 2026) read those cases together and held tower dumps per se unconstitutional. The court rejected the government’s two-or-more-towers limit on the ground that the carrier still has to search every record to find the overlap, so the narrowing comes after the search rather than before it.
The Mississippi order binds no other court, and the case has no defendant and no obvious route to appeal, so it may stand as one opinion for some time. Arnold & Porter’s write-up notes that defense counsel will cite it in suppression motions and that prosecutors will answer with the good-faith exception for warrants issued elsewhere. Either way, the question the judge asked is now the question every tower dump will face: did the search narrow the records, or did the analysis?
Narrowing the list without losing the record
In most cases the answer is the analysis. Under the Mississippi ruling, that alone cannot save the warrant. Everywhere tower dumps are still issued, it makes the analysis the thing that has to be defensible, and it has to produce two things at once: a short list of devices that matter, and a complete, unaltered record of everything that was returned, including the devices that were cleared.
The spreadsheet route is six tabs and a COUNTIF across them. It works until someone asks which sector a phone hit at the third scene, or whether the match survives once the UTC and local times are reconciled, or how many innocent devices were in the return in the first place.
In SentVi each tower dump is imported as its own saved Visualizer: one entity per device, one entity per tower with its coordinates, and one relationship for each connection with the timestamp attached. Every tower with latitude and longitude lands on the Map as soon as the import finishes. From there:
- Which devices appear at more than one scene? Select the saved Visualizers for the six locations and run Multi-Visualizer Analysis. It returns the entities that appear in several of them. Against thousands of rows per scene, that is usually a very short list, and it is the only list that leaves the room.
- Did they hit the right sector at the right time? The Time Range slider narrows the diagram to the window that matters, down to one second, and Geo Query draws the circle around the scene so only the towers and devices inside it stay on the chart.
- Can the full return be shown to the court? Yes, because nothing was deleted to get the short list. The complete dump sits in the database, and the chart is a view of it, not a replacement for it. If a judge or opposing counsel asks how many devices were in the return and how many were cleared, that is a count, not a reconstruction.
- Who sees the innocent records? Only the analyst. The full dataset stays with the analyst, and only the devices that appear across scenes are passed along, so bystanders are never named in what investigators or counsel receive. Whether that satisfies a court depends on the warrant, not the software.
For defense counsel the same chart answers the other question. If the state’s case rests on a phone that appeared at two of six towers, the chart shows the sector, the time, and the size of the wedge at each one, and it shows how many other phones hit the same towers in the same windows. A match that looks decisive in a spreadsheet can look like one of forty on a map.
What the record should be able to say
A tower dump that holds up, on either side of the courtroom, can answer four questions without anyone opening a new file: how many devices came back, how many appeared at more than one scene, what the time and sector were for each of those, and what happened to everyone else. If the analysis can produce that on demand, the argument is about the warrant. If it cannot, the argument is about the analysis, and that is the one that loses.
If you have a tower dump return in your case, or six of them, book a demo and bring them. We will run the cross-reference while you watch.
