Opinion
8:21-cr-309-VMC-MRM
02-28-2023
UNITED STATES OF AMERICA v. SCOTT CARPENTER, JR.
REPORT AND RECOMMENDATION
MAC R. MCCOY UNITED STATES MAGISTRATE JUDGE
Pending before the Court is Defendant's Motion to Suppress Evidence From a Geofence Warrant and Request for Evidentiary Hearing, filed on August 1, 2022. (Doc. 93). Defendant, Scott Carpenter, Jr., is charged with one count of Conspiracy to Interfere with Commerce by Robbery (18 U.S.C. § 1951(a)) (Count One); four counts of Interference with Commerce by Robbery (18 U.S.C. § 1951(a)) (Counts Two, Four, Six, Eight); and four counts of Brandishing a Firearm During or in Relation to a Crime of Violence (18 U.S.C. § 924(c)) (Counts Three, Five, Seven, Nine). (Doc. 30 at 1-7).Defendant seeks “to suppress the geofence warrant, its returns, and their fruits, to include any additional warrants for Mr. Carpenter's Google account, cell site warrants, warrants to search his home and car, and a DNA warrant.” (Doc. 93 at 24).
Pinpoint page citations for documents refer to CM/ECF pagination.
The Government filed a response in opposition on September 6, 2022. (Doc. 106). Defendant filed a reply on October 7, 2022. (Doc. 112). The Undersigned conducted an evidentiary hearing that was initially set for December 14, 2022 (Doc. 135) but was ultimately continued to January 3, 2023 due to technology failures in the courtroom (Doc. 140). Following the evidentiary hearing, both parties submitted supplemental briefing. (Docs. 143, 148). This matter is ripe for review.
A transcript of the evidentiary hearing is filed at Doc. 150. The Undersigned refers to the transcript herein as “Tr.” followed by the appropriate page number.
For the reasons explained below, the Undersigned recommends that Defendant's Motion to Suppress Evidence From a Geofence Warrant and Request for Evidentiary Hearing (Doc. 93) be DENIED.
I. Background
A. Robberies
Defendant is charged with one count of conspiracy to interfere with commerce by robbery, four counts of interference with commerce by robbery, and four counts of brandishing a firearm during or in relation to a crime of violence. (See Doc. 30). In January 2021, Bureau of Alcohol, Tobacco, Firearms and Explosives (“ATF”) “agents were investigating a series of eleven armed robberies and attempted armed robberies that had occurred in the greater Tampa Bay area between August 12, 2018, and December 8, 2020.” (Doc. 106 at 2; see generally Doc. 1). These robberies all concerned employees of Dollar Tree stores. (Doc. 30). Defendant has been charged in connection with four of these robberies: (1) the October 28, 2020 robbery in Brandon, Florida (“Brandon Robbery”) (Counts Two and Three); (2) the November 15, 2020 attempted robbery in Lakeland, Florida (“Lakeland Attempted Robbery”) (Counts Four and Five); (3) the November 23, 2020 robbery in Zephyrhills, Florida (“Zephyrhills Robbery”) (Counts Six and Seven); and (4) the December 8, 2020 robbery in Riverview, Florida (“Riverview Robbery”) (Counts Eight and Nine). To provide context for the geofence warrant at issue, the Undersigned provides summaries of the events below, drawn from the Superseding Indictment (Doc. 30), the Complaint (Doc. 1), and the parties' briefing (Docs. 93, 106, 112).
1. Brandon Robbery
On October 28, 2020, an armed robbery occurred around 9:28 p.m. in the parking lot of the Dollar Tree store located at 509 E. Brandon Boulevard, Brandon, Florida. (Doc. 1 at 10; see also Doc. 30 at 2). Two of the store's employees walked to their cars to travel to a bank to deposit $1,204.00 in store proceeds contained in a money bag. (Doc. 1 at 11). The robber approached them, pointing a black handgun, and told them to “give me the deposit money.” (Id.). The employee holding the money bag threw the bag to the ground, after which the robber picked up the bag and fled the scene on foot. (Id.). Both employees described the robber as a Black male, 5' 7” tall, between the ages of 20-30 years old, approximately 140 pounds, and wearing a black hoodie and green pants. (Id. at 11-12).
Defendant's motion cites the Hillsborough County Sheriff's Office (“HCSO”) police reports for additional details, but a copy of that report has not been provided. (Doc. 93 at 2). Based on Defendant's representations of that report, “HCSO [officers] collected a green ‘Surge' can at the scene, but it yielded no fingerprints.” (Id.). HCSO officers also reviewed camera footage from both the bank and nearby businesses. (Id.). Finally, “[s]ubpoenas were issued to T-Mobile and Sprint for two phone numbers that were redacted in the HCSO report, but a search of those records yielded no results.” (Id.).
2. Lakeland Attempted Robbery
On November 15, 2020, an employee of the Dollar Tree store located at 3629 S. Florida Avenue, Lakeland, Florida experienced an attempted armed robbery. (Doc. 1 at 12; see also Doc. 30 at 3-4). Around 10:36 p.m., the employee closed the store and went to go deposit store funds in the night deposit box at Fifth-Third Bank located at 4421 S. Florida Avenue. (Doc. 1 at 12). When the employee arrived at the bank and pulled into the deposit drive-thru lane, a male approached her vehicle and banged on the rear passenger window. (Id.). The male brandished a black handgun and pointed it at the employee. (Id. at 12-13). Subsequently, he tapped the gun on the driver's side window while trying to open the door. (Id. at 13). The employee drove away and called the police. (Id.).
Defendant's motion cites the Lakeland Police Department's police report for additional details, but a copy of that report has not been provided. (Doc. 93 at 3). Based on Defendant's representations of that report, the employee described the suspect as a “heavyset male, 5'9, black ski mask, and hooded jacket.” (Id.).
3. Zephyrhills Robbery
On November 23, 2020, an employee and a co-worker of the Dollar Tree Store located at 7749 Gall Boulevard, Zephyrhills, Florida closed the store and traveled to the Wells Fargo Bank located at 7780 Gall Boulevard in Zephyrhills around 10:40 p.m. (Doc. 1 at 13; see also Doc. 30 at 4-5). The pair intended to deposit $2,920.05 in store funds in the night deposit box at the bank. (Doc. 1 at 13). As the employee walked from her vehicle, “a heavy-set Black male, approximately 5' 4” tall, who was brandishing a pistol and wearing a ski mask, black jacket and pants” approached her. (Id.). The robber demanded the money, wrapped his arms around the employee, and threw her to the ground. (Id.). The co-worker then alerted the employee to the robber's gun. (Id.). The employee surrendered the money bag, and the robber took it and fled the area. (Id.). Both the employee and the co-worker then heard the screeching of car tires. (Id.).
Defendant's motion cites the Zephyrhills Police's report for additional details, but a copy of that report has not been provided. (Doc. 93 at 3). Based on Defendant's representations of that report, one employee described the suspect as follows: “5'5, Black male, puffy jacket, black jeans, black shoes, [and] ski mask.” (Id.). The other described the suspect “as 5'4 and 240 pounds.” (Id.). Defendant also mentions that “[b]ank video surveillance showed a suspect in dark clothing and a hoodie, but the person's face was not visible.” (Id.).
4. Riverview Robbery
On December 8, 2020, an employee of the Dollar Tree store located at 10680 E. Bay Road, Gibsonton, Florida and a co-worker traveled to the Fifth-Third Bank located at 10417 Gibsonton Drive in Riverview around 10:56 p.m. to deposit $1,475.81 of store funds in the night deposit box. (Doc. 1 at 13-14; see also Doc. 30 at 6). Around twenty-two seconds after arriving at the night deposit box, “a Black male, approximately 5'6” tall, brandishing a black handgun and wearing all-black clothing” approached the employee's car and reached into it. (Doc. 1 at 14). The employee moved to the front passenger seat of the vehicle while the robber searched for the money. (Id.). The robber eventually found the money, grabbed it, and ran from the vehicle. (Id.).
Defendant's motion cites the HCSO police report for additional details, but a copy of that report has not been provided. (Doc. 93 at 4). Defendant makes the following representations about that report: “HCSO reports stated that one employee described the suspect as a Black male, black nightcap, black mask, black jacket, and black pants. The other described him as 5'6, 140-160 pounds, 30-40 years old. Neither employee could see the suspect's face. Bank video showed a suspect in dark clothing and hoodie, but his face was not visible. A nearby Walmart video surveillance showed what could be the suspect's car. The HCSO report stated no fingerprints of value were found on the Dollar Tree employee's car.” (Id.).
B. Geofence Warrant
1. Application
On January 22, 2021, the ATF applied for and received a geofence warrant. (See Doc. 106-1). This type of warrant “is what has been termed a ‘reverse-location' warrant: the perpetrator of the crime being unknown to law enforcement, the warrant identifies the geographic location where criminal activity happened and seeks to identify cell phone users at that location when the crime occurred.” Matter of Search of Info. That is Stored at Premises Controlled by Google LLC, 579 F.Supp.3d 62, 69 (D.D.C. 2021) (hereafter, “D.C.”). In summary, the geofence warrant directed to Google works as follows:
Phones with Google OS and Google applications communicate location information to Google through use of common phone applications (e.g., YouTube and Google Maps) and other functions, including use of wi-fi, Bluetooth, and GPS. That location information is then associated with the device, and the device can be traced back to a particular account and accountholder who has “signed in” on the device. The search warrant at issue
ultimately seeks, through a multiple-step process, information that can be used to identify the accountholder associated with devices that Google has located within a certain geographic area, and during a specific time window, set by the government.Id. at 71-72.
In this case, Special Agent Jeffrey Burt's (“SA Burt”) affidavit in support of the warrant detailed each of the eleven armed robberies being investigated, including a description of the location and time of the robberies. (Doc. 106-1 at 10-20). The affidavit noted apparent similarities between the robberies, such as the robber wearing black clothing, being commonly described as between 5'7” and 5'9” tall with a stocky build, and approaching employees of Dollar Tree stores at night as they attempt to deposit store funds. (Id. at 20-21). The affidavit also asserted that “[t]hree different possible suspect vehicles have been identified, indicating the suspect may be working with co-conspirator(s) during the robberies.” (Id. at 21). In connection with the Brandon Robbery, the affidavit mentioned a video showing “a bright light in the robber's hands” that SA Burt believed “likely came from the screen of a cellular telephone being used by the robber at the time.” (Id. at 18). Finally, the affidavit “explained how Google services work, how Google's services are linked to users' email accounts (which are usually Google ‘Gmail' accounts) and cell phones; how Google's services obtain location data from cell phones on which those services are enabled; and how ubiquitous such cell phones are.” (Doc. 106 at 3 (citing Doc. 1061 at 3-9)).
Based on these representations, SA Burt requested that the Court authorize a search of Google's information “to determine what Google-connected devices, if any, were in the target areas.” (Doc. 106-1 at 21). The requested search parameters included a thirty-minute window (fifteen minutes before and fifteen minutes after the alleged robberies) and a geographic circle with a radius of 800 meters centered around particular coordinates (capturing the location of the robberies and a roughly half-mile radius around each one). (Id. at 21-22, 25-27). Although there were eleven events at issue, the warrant only sought information as to ten because the Lakeland Police Department had secured a search warrant for one of the robberies. (Id. at 22 n.11).
To allow Google to provide the requested information, SA Burt suggested a two-step process. (Id. at 22). First, Google would query its records and “produce to the government anonymized information specifying the corresponding unique device ID, timestamp, coordinates, display radius, and data source” for each device reported within the target locations during the relevant time periods. (Id. at 22, 2829). Attachment A to the warrant contained this requested information. (Id. 25-27). After Google produces that anonymized list, “the government will then review this list in order to prioritize the devices about which it wishes to obtain associated information.” (Id.). Second, “Google will then be required to disclose to the government the information identifying the Google account(s) for those devices about which the government further inquires.” (Id. at 23). Attachment B to the warrant outlined the warrant's two-step process. (Id. at 28-29). In essence, this two- step process would allow SA Burt “to cross-reference the data from each incident against the data from the other incidents, so as to identify any common Google accounts that appear at more than [one] incident.” (Id. at 4). For those common accounts, Google would then “reveal the subscriber information for that account, so that [SA Burt] may identify potential suspects, trigger-pullers, witnesses, and/or victims by their presence at multiple incidents.” (Id.).
2. Production
United States Magistrate Judge Sean P. Flynn found probable cause and authorized the geofence warrant in this case. (Id. at 1). Based on the representations in the Government's response, SA Burt served the warrant, and Google provided anonymized location history data for the ten locations, which contained approximately 1,040 unique email accounts. (Doc. 106 at 4-5). SA Burt then asked Google to provide the identifying data for only two accounts. (Id.; see also 106-2). As relevant here, Google revealed that one of the email accounts was scarpenter471@gmail.com. (Doc. 106 at 5). “Subscriber information for that account indicated that it belonged to the defendant.” (Id.). From this information, SA Burt authored two additional affidavits for warrants related to the scarpenter471@gmail.com account. (Id.). Judge Flynn authorized the first, 8:21-mj-1512-SPF, and United States Magistrate Judge Anthony E. Porcelli authorized the second, 8:21-mj-1858-AEP. (Id.).
II. Summary of the Evidentiary Hearing
At the January 3, 2023 evidentiary hearing, there were no objections to any party's exhibits. (Tr. at 8:4-9). Accordingly, all exhibits as listed on the parties' exhibit lists filed at Doc. 132 and 134 were moved into evidence. (Id. at 8:12-17). As for witnesses, each party called only one: (1) Defendant called Spencer J. McInvaille, his expert on mobile devices and digital forensics, (Doc. 133); and (2) the Government called SA Burt, (Doc. 131). The Undersigned will address, as relevant to the adjudication of Defendant's motion, the testimony and related evidence of each witness below.
A. Mr. McInvaille
1. Direct
Currently employed with Envista Forensics as a technical lead, Mr. McInvaille provides consulting and analytical services to attorneys in the practice of mobile device forensics. (Def. Ex. 1, Doc. 140-11 at 2). He has “extensive training and experience analyzing location data such as call detail records, global positioning data, mobile device forensics, mobile networks, wireless communications, and rendering opinions about these data types.” (Id.; see also Tr. at 12:5-11). He has been qualified and testified as an expert over thirty-five times in state and federal courts. (Doc. 140-11 at 2). With no objection from the Government, Defendant moved to qualify Mr. McInvaille as an expert witness in the field of mobile device and digital forensics, and the Court so qualified him. (Tr. at 19:7-13).
Mr. McInvaille testified that Google is a company that provides various services, applications, operating systems, and other ventures. (Id. at 19:25-20:2). As examples, Mr. McInvaille stated that Google Maps is an application and Android is a mobile operating system provided by Google. (Id. at 20:12-17). Mr. McInvaille testified that “location data” is an umbrella term used to refer to all types of data that can be used to locate a person. (Id. at 20:18-21:2). Location data can be gathered directly by a device or by the mobile device carrier. (Id. at 21:3-10).
A subset of location data is Location History, “an actual product of Google.” (Id. at 22:17-22). Location History is “a service that a user can opt into and store their [sic] location data.” (Id. at 22:19-21). To collect Location History, Mr. McInvaille testified that Google allows devices using its Android operating system to survey the areas the devices are in to search for cell towers or Wi-Fi routers. (Id. at 23:24-24:1). The devices attempt to understand the positions of the towers or routers based on relative signal strength values. (Id. at 24:1-13). From that information, Google can estimate where the device is in relation to the relative distances between the towers or routers. (Id. at 24:21-24). Mr. McInvaille also testified that Location History can derive from Bluetooth and GPS, with Bluetooth being uncommon and GPS being very common. (Id. at 25:12-14).
Mr. McInvaille reiterated that Location History is a unique service to Google. (Id. at 27:21-22 (“[It] is the only company that has that service.”)). He differentiated Location History from call detail records in that call detail records are maintained by the company. (Id. at 28:4-15). In contrast, the user maintains control over Location History, including the ability to change, edit, and remove the information. (Id. at 28:12-15, 32:21-33:6). Although it was true at one point in time, Mr. McInvaille testified that the mere use of an Android operating system does not automatically turn on Location History. (Id. at 28:20-25). Now, Location History must be enabled either with the setup of a new device or at a later time through the device's settings. (Id. at 29:4-6). The use of Location History requires a Google Gmail account. (Id. at 29:23-25). Signing out of a Google Gmail account on a device or turning off the device will stop Location History from tracking that device. (Tr. at 31:6-8). Mr. McInvaille testified that only about a third of Google users enable Location History at any time. (Id. at 29:12-13).
As to geofence warrants, Mr. McInvaille testified that those types of warrants are used to search for unknown users through location data collected on their devices to determine whether the users were in a particular location. (Id. at 34:5-7). A geofence warrant based on Location History will capture users of Location History in an area but only those users who have Location History enabled. (Id. at 35:17-21). Mr. McInvaille believes that geofence warrants began being used by law enforcement around 2015 or 2016, but his first experience of working on a geofence case was in 2018. (Id. at 35:25-36:5).
Mr. McInvaille testified that he most commonly sees a three-step process with geofence warrants. (Id. at 38:13-14). At step one, a geofence is drawn on a map in any shape, and date and time parameters are applied to that geographic region. (Id. at 38:17-23). Google then searches all of its accounts to determine which individual users' Location History data points are responsive to the geofence. (Id. at 39:1-4). After, Google provides an anonymized list of users that includes an identifying device number, date and time measurements, estimated latitude and longitude, the source of the data (e.g., GPS, Wi-Fi, cell tower, etc.), and the display radius of the circle drawn around the location data. (Id. at 39:21-40:10). The display radius is based on Google's confidence in its approximation of the location with roughly 68 percent accuracy. (Id. at 45:7-12, 61:18-24).
When determining the users at step one, Mr. McInvaille testified that Google can use three separate methods. (Id. at 40:21-22). The most common one is that the center of Google's display radius circle falls somewhere within the bounds of the geofence. (Id. at 40:22-41:3; see also Def. Ex. 6, Doc. 140-17 (Example 1)). Another is that both the center of Google's circle and the radius of that circle fall within the bounds of the geofence. (Tr. at 41:4-10; see also Doc. 140-17 (Example 3)). The third method is if any point within Google's circle (irrespective of where the center point is) intersects with the geofence. (Tr. at 43:24-44:2; see also Doc. 140-17 (Example 2)). Mr. McInvaille has seen warrants specify which of these three methods Google should use for finding individual users responsive to the warrant. (Tr. at 43:11-13).
Turning to the second step in the common three-step geofence warrant process, Mr. McInvaille testified that “you can select to see more information about those users” on the anonymized list. (Id. at 45:19-22). He labels this request for more information as a request for “contextual data” because it provides additional information on the devices as they travel through or remain in the selected area by removing the geofence and expanding the time frame. (Id. at 45:22-46:7). Finally, at step three, Mr. McInvaille testified that a request for subscriber information is made. (Id. at 47:3-8).
Specific to the warrant issued in this case, Mr. McInvaille testified that a two-step process was used. (Id. at 51:7-13). At step one in this case, Mr. McInvaille states that Google would have to search all of its Location History users for each of the ten requested geofences. (Id. at 52:8-10, 52:21-22). There would then be ten anonymized lists created, one for each of the geofences. (Id. at 55:5-9). After receiving those lists, the second step involved getting the subscriber information. (Id. at 55:16-20). Mr. McInvaille noted that the warrant included no factors for prioritizing what subscribers would be chosen at step two. (Id. at 56:14-21). In all, the produced lists included approximately 1,040 users with around 5,200 data points. (Id. at 62:10-11). But the subscriber information for only two users was requested. (Id. at 68:8-11).
Defendant's data was one of those two users. (Id. at 69:15). His data was found within two of the robbery locations. (Id. at 68:16-17, 69:20-22). In reviewing the data, though, Mr. McInvaille found that there were other users who appeared at multiple robbery locations but whose subscriber information was not sought by the Government. (Id. at 71:23-72:1).
Mr. McInvaille testified about the specific geofences drawn at each of the four robberies at issue for Defendant: the Brandon Robbery, (id. at 75:15-81:4); the Lakeland Attempted Robbery, (id. at 81:9-83:15); the Zephyrhills Robbery, (id. at 83:20-85:14); and the Riverview Robbery, (id. at 85:21-88:10). He noted that the geofences included restaurants and churches (Id. at 76:15-19), residential areas and doctors' offices (id. at 81:24-25), and banks and stores (Id. at 84:11-13). Finally, Mr. McInvaille testified that when he created his own maps of the geofences and the data, he had to reduce the geofence radius because the data was more concentrated than what he would normally see. (Id. at 78:4-15, 88:19-90:1).
2. Cross-Examination
On cross-examination, Mr. McInvaille testified that every time a mobile phone reaches out and pings a cell tower, a record is generated at the service provider. (Id. at 106:11-21). A different kind of record is generated every time there is activity on the phone, such as making a call or sending a text message. (Id. at 106:22-107:4).
Mr. McInvaille also testified that Location History is provided by Google, at least in part, for “enhancing [the] user experience.” (Id. at 14:10-17, 115:6-11). For example, Location History can be used to learn about a user's routines to suggest potentially helpful or desired products and services. (Id. at 115:9-11 (“[I]t says hey, do you want to stop at Starbucks, because you're one of the people that stops at Starbucks all the time.”)). In other words, Location History is a distinct service from simply using a phone or application to get directions to a particular location. (Id. at 116:12-21).
Mr. McInvaille testified that other companies, such as Apple, also collect location data about their users. (Id. at 119:20-22). Apple, for instance, collects the location data right to the device, unlike Google's Location History which stores data on a cloud. (Id. at 119:20-120:). For that reason, Mr. McInvaille testified that “you don't see Apple geofence warrants.” (Id. at 119:21).
When asked about step two in the three-step warrant process, Mr. McInvaille testified that he was unaware of any limit on the number of people for whom contextual data could be requested. (Id. at 140:20-141:1). But he further testified that he has reviewed other geofence warrants that limit the expanded time frame requested for the contextual data. (Tr. at 141:25-142:1). Although those limitations are not generally found in the warrants themselves, Mr. McInvaille stated that Google requests that time limitations be imposed. (Id. at 142:5-8, 142:18-21).
Mr. McInvaille also testified on re-direct, but that testimony elicited no new and relevant information beyond what has already been described. (Tr. at 158:23168:15).
B. SA Burt
1. Direct
On direct, SA Burt testified that he is a special agent with the ATF in Tampa, where he has been for eighteen years. (Id. at 169:25-170:1). He has prior law enforcement experience working on armed robberies, felon in possession cases, undercover operations, and armed drug trafficking. (Id. at 170:1-6). He became involved in this case when local police departments requested assistance. (Id. at 170:12-19).
SA Burt first became aware of Google geofence warrants around 2017 or 2018. (Id. at 171:12). He has received no formal training from the ATF on Google geofence warrants. (Id. at 171:21-23). But he received informal training, which he defined as having conversations with others with experience on geofence warrants, including technical advisors. (Id. at 172:9-13). SA Burt also spoke with a former law enforcement partner, Agent Chad Horst, about Agent Horst's use of a geofence warrant to solve a case in Atlanta. (Id. at 172:17-25). Agent Horst sent SA Burt a copy of the warrant he drafted in his case, and SA Burt reviewed that warrant in preparing the warrant in this case. (Id. at 173:15-19, 174:6-8; see also Gov't Ex. 1, Doc. 140-5). When Agent Horst discussed his warrant with SA Burt, he expressed no concerns about the legality of the warrant, or that he received any pushback from a local prosecutor or judge about the warrant. (Tr. at 175:1-21). Beyond Agent Horst's experience, SA Burt was aware of another case in Tampa that involved the use of a geofence warrant, but he was not the author of that warrant. (Id. at 176:411). He did, though, review that warrant as a possible example to use in this case. (Id. at 179:22-180:6; see also Gov't Ex. 2, Doc. 140-6).
SA Burt testified that when he requested the geofence warrant, police had executed a tower dump search warrant. (Id. at 178:10-13). He testified that two criminal analysts helped review and analyze the data from that tower dump: Yesenia Toledo and Elizabeth Capitano. (Id. at 179:8-14).
Ms. Capitano was previously Ms. Mendenhall before getting married. (Tr. at 179:15-16).
When SA Burt was preparing the warrant in this case, he initially drafted his affidavit before consulting AUSA Michael M. Gordon (the AUSA on this case). (Id. at 180:10-21). SA Burt and AUSA Gordon edited the affidavit together and sent multiple drafts back and forth. (Id. at 180:25-181:8). During his conversations with AUSA Gordon, SA Burt testified that there was never a concern about the legality of a geofence warrant, that other judges had rejected similar warrants, or that submitting the warrant to the judge would be like a “roll [of] the dice.” (Id. at 181:917).
After finishing the affidavit and the warrant, SA Burt submitted the warrant to Judge Flynn. (Id. at 182:8-9). Because of the COVID pandemic, the signing of the warrant occurred over the phone, in which Judge Flynn, SA Burt, and AUSA Gordon all participated. (Id. at 182:9-25). SA Burt testified that Judge Flynn expressed no concerns or reservations about the warrant. (Id. at 183:4-14).
After SA Burt submitted the warrant to Google, he testified that a Google employee named “Tom” called him. (Id. at 186:25-187:8). Tom told SA Burt that the search warrant was generating a large amount of data, and he questioned SA Burt about the possibility of narrowing the scope of the search. (Id. at 187:16-19). SA Burt agreed to narrow the search parameters. (Id. at 188:13-15). SA Burt emailed Tom to reduce the search area of each location to a range of 200 to 400 meters depending on the location. (Id. at 188:25-189:17; see also Gov't Ex. 4, Doc. 140-8 at 4). While shrinking the size of the geofence, SA Burt still intended for the search to capture any get-away drivers. (Tr. at 189:8-11, 190:20-191:1).
Once the search was narrowed, Google sent SA Burt the spreadsheets containing the anonymized list of users found within the geofence. (Id. at 191:2125). SA Burt then sent the spreadsheets to Ms. Capitano and Ms. Toledo. (Id. at 192:5-8). After analyzing the data, Ms. Capitano returned to SA Burt with a recommendation that subscriber information be requested as to two devices. (Id. at 194:6-10, 198:21-25). SA Burt did not question her about that decision because he “trusted her judgment.” (Id. at 195:2-9). SA Burt then requested the subscriber information for those two users from Google. (Id. at 195:10-14). When SA Burt initially emailed Google, however, he only included a request for one device. (Id. at 195:23-196:1; see also Doc. 140-8 at 6). He did this because Ms. Capitano was most interested in the results of that device. (Tr. at 195:25-196:3). SA Burt did not include the other device until a later correspondence with Google. (Id. at 198:21-25; see also Doc. 140-8 at 8). Finally, SA Burt testified that he does not recall why he decided to use a two-step process here rather than a three-step one. (Tr. at 196:24-197:2).
2. Cross-Examination
On cross-examination, SA Burt testified that that he has had no training on GPS, Wi-Fi networks, Bluetooth, Google, location data, Location History, Android, or geofences but has had some training on cell towers, cell phones, and cell phone extractions. (Id. at 209:14-210:25). SA Burt also testified that he did not consult either Ms. Toledo or Ms. Capitano with respect to the drafting of the geofence warrant. (Id. at 219:3-6). Regarding paragraphs 9 to 19 of the affidavit concerning his training and experience, SA Burt testified that he relied more on experience rather than training. (Id. at 219:22-223:2). Finally, SA Burt agreed that “law enforcement has free reign to prioritize from step one to step two” based on the warrant in this case. (Id. At 227:7-10).
The Government conducted no redirect examination of SA Burt.
III. Analysis
The parties' arguments concern three general issues: (1) whether the geofence warrant is constitutionally valid; (2) whether the good-faith exception applies; and (3) whether Defendant has standing to challenge the warrant. (Docs. 93, 106, 112, 143, 148).For the reasons below, the Undersigned finds that the good-faith exception applies. Therefore, in the interest of judicial economy and efficiency, the Undersigned finds that the Court need not address the standing or validity issues because, even assuming arguendo that Defendant has standing and that the warrant is invalid, the Undersigned finds that suppression is not warranted.
In his supplemental brief, Defendant also argues that both the Fourth Amendment and the Stored Communications Act, 18 U.S.C. § 2701 et seq., require the Government to obtain a warrant for Location History. (Doc. 143 at 1-2). The Government holds the contrary view: Location History “may be obtained via a court order pursuant to 18 U.S.C. § 2703(d).” (Doc. 148 at 1). Although Mr. McInvaille provided some testimony on this topic, (Tr. at 33:18-25, 161:25-162:14), the Court need not address this issue because doing so would constitute an advisory opinion. A § 2703(d) order is not at issue here as the Government obtained a search warrant instead. Accordingly, the Undersigned addresses only the parties' arguments related to the warrant.
Standing in the Fourth Amendment context is “shorthand for capturing the idea that a person must have a cognizable Fourth Amendment interest in the place search before seeking relief for an unconstitutional search; but it should not be confused with Article III standing, which is jurisdictional and must be assessed before reaching the merits.” Byrd v. United States, 138 S.Ct. 1518, 1530 (2018). In other words, the question of standing in this case “is not a jurisdictional question and hence need not be addressed before addressing other aspects of the merits of a Fourth Amendment claim.” Id.
A. Parties' Arguments Regarding the Good-Faith Exception
The Government maintains that the Court need not reach the merits of Defendant's constitutional arguments because this case falls under the good-faith exception. (Doc. 106 at 6-7). The Government contends that SA Burt “acted in good faith at every stage.” (Id. at 8). Specifically, the Government points to SA Burt consulting a federal prosecutor to approve the sufficiency and constitutionality of the affidavit and the obtaining of authorization for the warrant from Judge Flynn as the grounds for his good faith. (Id. at 8-9).
Defendant urges the Court to find the good-faith exception inapplicable.\ (Doc. 143 at 3). He first argues that SA Burt was a “novice in dealing with geofence warrants.” (Id.). Defendant references the fact that SA Burt received no training on geofence warrants. (Id. at 4). This lack of training conflicts, Defendant asserts, with the affidavit submitted with the warrant in which SA Burt relied on his “training and experience.” (Id.). Moreover, Defendant finds fault in SA Burt initially deciding on an 800-meter radius to search but then reducing it based on a Google employee telling him that the geofence areas should be smaller. (Id. at 5).
In its supplemental brief, the Government characterizes Defendant's arguments as “unfair.” (Doc. 148 at 4). The Government asserts that SA Burt never claimed to be an expert on cellular technology or Google, but that Defendant “seeks to hold SA Burt to the standard of an expert witness when no such level of knowledge is required to state the basic facts contained in the affidavit.” (Id.). Further, the Government argues that even if Defendant is correct that SA Burt claimed to know technical facts that he did not, the good-faith exception still applies because there has been no showing that Judge Flynn was misled by information in the affidavit. (Id. at 5). And the Government contends that SA Burt's decision to reduce the scope of the search had no bearing on whether SA Burt acted in good faith. (Id. at 8). Finally, the Government argues that Defendant is attempting to create a new definition of good faith that is inconsistent with existing precedent. (Id. at 9).
B. Whether the Good-Faith Exception Applies
“Under what is known as the ‘exclusionary rule,' evidence seized as a result of an illegal search may not be used by the government in a subsequent criminal prosecution.” United States v. Martin, 297 F.3d 1308, 1312 (11th Cir. 2002). The exclusionary rule “is a judicially created remedy designed to safeguard Fourth Amendment rights generally through its deterrent effect, rather than a personal constitutional right of the party aggrieved.” United States v. Calandra, 414 U.S. 338, 348 (1974). “The exclusionary rule serves to deter police misconduct by preventing the introduction of evidence obtained through police illegality.” United States v. Virden, 488 F.3d 1317, 1322 (11th Cir. 2007) (citing Nix v. Williams, 467 U.S. 431, 442-43 (1984)).
The Supreme Court created a “good-faith exception” to the exclusionary rule in United States v. Leon, 468 U.S. 897 (1984). The exception applies to evidence that otherwise would be prohibited by Fourth Amendment violations when the evidence is obtained “by officers reasonably relying on a warrant issued by a detached and neutral magistrate” that is later invalidated for lack of probable cause. Id. In other words, “[w]hen law enforcement officers act in good faith and in a reasonable reliance upon a judge's order, exclusion is not warranted because there is no unlawful conduct to deter.” United States v. Goldstein, 989 F.3d 1178, 1196 (11th Cir. 2021). Courts may consider the totality of the circumstances by “look[ing] beyond the four corners of the affidavit and search warrant to determine whether [the officer] reasonably relied upon the warrant.” Martin, 297 F.3d at 1318. “The government bears the burden of demonstrating that the good faith exception applies.” United States v. Morales, 987 F.3d 966, 974 (11th Cir. 2021).
“The Leon good faith exception applies in all but four limited sets of circumstances.” Martin, 297 F.3d at 1313. Those four scenarios in which the goodfaith exception does not apply are as follows:
(1) where “the magistrate or judge in issuing a warrant was misled by information in an affidavit that the affiant knew was false or would have known was false except for his reckless disregard of the truth”; (2) “where the issuing magistrate wholly abandoned his judicial role in the manner condemned in” Lo-Ji Sales, Inc. v. New York, 442 U.S. 319, 99 S.Ct. 2319, 60 L.Ed.2d 920 (1979); (3) where the affidavit supporting the warrant is “so lacking in indicia of probable cause as to render official belief in its existence entirely unreasonable”; and (4) where, depending upon the circumstances of the particular case, a warrant is “so facially deficient-i.e., in failing to particularize the place to be searched or the things to be seized-that the executing officers cannot reasonably presume it to be valid.”Id. at 1313 (quoting Leon, 468 U.S. at 923).
Defendant does not specifically articulate which of these four he asserts apply to his case. (Docs. 93, 112, 143). Construing Defendant's arguments broadly, the Undersigned finds that Defendant's argument falls under the first scenario-namely the good-faith exception does not apply because SA Burt misled Judge Flynn. Defendant also distinguishes his case from United States v. Chatrie, 590 F.Supp. 3D 901 (E.D. Va. 2022) as evidence of why the good-faith exception should not apply. The Undersigned addresses both of Defendant's arguments in turn.
To the extent Defendant intended to rely on the second, third, or fourth circumstances, he has not properly presented those arguments to the Court or otherwise supported them with appropriate legal authority and argument. Thus, the Undersigned finds that Defendant has waived argument under those exceptions. Cf. Center v. Sec'y, Dep't Homeland Sec., 895 F.3d 1295, 1299 (11th Cir. 2018) (“[W]e have repeatedly explained that arguments briefed in the most cursory fashion are waived.”); Stanley v. City of Florida, No. 6:20-cv-629-WWB-GJK, 2021 WL 6333059, at *4 (M.D. Fla. Mar. 1, 2021) (“Conclusory, vague, and unsupported arguments are deemed waived and will not be considered by this Court.”). Even if Defendant's arguments could be liberally construed to include the third and fourth scenarios, the Undersigned finds that they are inapplicable. In short, whatever validity issues exist with the warrant, the warrant was not so lacking in indicia of probable cause or so facially deficient because the legality of geofence warrants presents tough legal questions with relatively few judicial opinions to illuminate answers. See United States v. Blake, 868 F.3d 960, 975 (11th Cir. 2017) (“And while the warrants may have violated the particularity requirement, whether they did is not an open and shut matter; it is a close enough question that the warrants were not ‘so facially deficient' that the FBI agents who executed them could not have reasonably believed them to be valid.”); see also United States v. McLamb, 880 F.3d 685, 691 (4th Cir. 2018) (declining to suppress evidence when “the legality of an investigate technique is unclear and law enforcement seeks advice from counsel before applying for the warrant”).
1. Misled
Defendant primarily argues that SA Burt misled Judge Flynn. (Doc. 143 at 7 (“Agent Burt misrepresented his own training and experience . . . which the magistrate judge relied on and trusted in his granting of the warrant.”). Specifically, Defendant claims that SA Burt misrepresented his training on the affidavit. SA Burt testified that that he had no training on GPS, Wi-Fi networks, Bluetooth, Google, location data, Location History, Android, or geofences but had some training on cell towers, cell phones, and cell phone extractions. (Tr. at 209:14-210:25). Yet the affidavit included representations that SA Burt relied on both his training and experience. (See, e.g., Doc. 106-1 at 4 ¶ 11 (“Based on my training and experience, I also know that many cellular devices feature Bluetooth functionality.”), 7 ¶ 17 (“Based on my training and experience, I know that Google collects and retains location data from devices running the Android operating system when the user has enabled Google location services.”)). As for those “training and experience” paragraphs in the affidavit, SA Burt testified that he relied more on experience than training. (Tr. at 219:22-223:2).
The Undersigned finds that SA Burt's liberal use of “training” in the affidavit was boilerplate at best and negligent at worst. For instance, although he admits having no formal training on location data or Google, SA Burt made several declarations about both based on his training and experience. (Doc. 106-1 at 5-9 ¶¶ 13-21). But the Undersigned does not find that the record establishes that Judge Flynn “was misled by information” in the affidavit that SA Burt “knew was false or would have known was false except for his reckless disregard of the truth.” Martin, 297 F.3d at 1313. “The Leon good faith exception requires suppression ‘only if the officers were dishonest or reckless in preparing their affidavit or could not have harbored an objectively reasonable belief in the existence of probable cause.'” Id. (citation omitted). SA Burt did have some training on cell phones generally. (Tr. at 209:14-210:1). He also includes within his definition of training those informal conversations he has had with others, including the technical advisors. (Id. at 172:913). And he relied more on his experience than training in making his affirmations. (Id. at 219:22-223:2). Crucially, Defendant does not argue that any of the information contained in those “training and experience” paragraphs was factually false or misleading. The Undersigned does not, therefore, find that SA Burt's affidavit was knowingly false or that he should have known about the falsity absent a reckless disregard of the truth. See Martin, 297 F.3d at 1320 (“The exclusionary rule is meant to guard against police officers who purposefully leave critical facts out of search warrant affidavits because these facts would not support a finding of probable cause.”).
2. Chatrie
Defendant does not specifically argue any other reason why the good-faith exception should not apply. (Docs. 93, 112, 143). He does, though, argue that SA Burt “is not in the same position as the officer in the Chatrie case.” (Doc. 143 at 8 (citing United States v. Chatrie, 590 F.Supp.3d 901 (E.D. Va. 2022)). In United States v. Chatrie, the court found that the Google geofence warrant was unconstitutional but refused to suppress the evidence because of the good-faith exception. 590 F.Supp.3d at 937. The court held that the warrant was not so lacking in indicia of probable cause as to render official belief in its existence entirely unreasonable for four reasons: (1) the agent consulted attorneys before submission of the warrant; (2) the agent obtained similar warrants in the past; (3) geofence warrants involve novel technology such that their legality is unclear; and (4) a magistrate judge had approved the warrant. Id. at 936-38.
Defendant attempts to distinguish his case from Chatrie by arguing that the agent in Chatrie “was able to rely on his experience in getting three other geofence warrants granted before the one he sought in Chatrie.” (Doc. 143 at 8). While Defendant is correct that SA Burt does not have experience authoring geofence warrants, he did speak to other agents about their experiences. (Tr. At 172:17-25). He reviewed the warrants that they had authored and obtained copies as examples to use in this case. (Id. at 173:15-19, 174:6-8, 179:22-180:6). Importantly, Defendant ignores that SA Burt, like the agent in Chatrie, consulted an attorney prior to applying for the subpoena. (Id. at 180:25-181:8). An agent's consultation with an attorney prior to seeking approval of a warrant is one indicator of objective good faith. United States v. Taxacher, 902 F.2d 867, 872 (11th Cir. 1990); see also United States v. Matthews, 12 F.4th 647, 656 (7th Cir. 2021) (“Consulting with the State's Attorney or similar prosecutorial officer certainly is one step a responsible and diligent officer can take, and such consultation is, in many respects, exactly what Leon's good-faith exception expects of law enforcement.”).
Nor does Defendant dispute that the legality of geofence warrants is a cuttingedge issue based on how recently geofence warrants have been used and challenged. See United States v. McLamb, 880 F.3d 685, 691 (4th Cir. 2018) (finding good-faith exception applicable when “the legality of an investigate technique is unclear and law enforcement seeks advice from counsel before applying for the warrant”). Finally, Judge Flynn authorized the warrant. While the Undersigned agrees with the court in Chatrie that “magistrate [judge] approval and consultation with the prosecution alone cannot and should not mechanically trigger the good-faith exception,” Chatrie, 590 F.Supp.3d at 938, the purpose of the exclusionary rule, as the court in Chatrie recognized, is to “‘meaningfully deter' improper law enforcement conduct.” Id. (quoting Herring v. United States, 555 U.S. 135, 144 (2009)).
Here, the Undersigned finds that when considering all the circumstances, the record does not suggest that SA Burt unreasonably relied on the warrant. After all, “‘an officer cannot be expected to question the magistrate's probable-cause determination' because ‘[i]t is the magistrate's responsibility to determine whether the officer's allegations establish probable cause and, if so, to issue a warrant comporting in form with the requirements of the Fourth Amendment.'” Messerschmidt v. Millender, 565 U.S. 535, 547 (2012) (quoting Leon, 468 U.S. at 921). Therefore, the Undersigned finds the good-faith exception applies, and Defendant's motion is due to be denied. See United States v. Davis, No. 2:21-CR-101-MHT-JTA, 2022 WL 3009240, at *9 (M.D. Ala. July 1, 2022), report and recommendation adopted, 2022 WL 3007744 (M.D. Ala. July 28, 2022) (“Because there is no law enforcement wrongdoing to deter, suppression is not warranted.”).
A more recent district court opinion on the use of a geofence warrant also found the good-faith exception applicable. United States v. Rhine, No. 21-0687 (RC), 2023 WL 372044, at *33 (D.D.C. Jan. 24, 2023). Although that court also found the warrant constitutionally valid, the court explained that it “would not suppress the evidence obtained from the Geofence Warrant under the good faith exception even if it had found that the [w]arrant lacked probable cause.” Id.
In sum, the Undersigned finds that the Government has met its burden of demonstrating that the good-faith exception applies in this case. As a result, the Undersigned finds that Defendant's motion to suppress is due to be denied.
Because the Undersigned finds that the good-faith exception applies, the Undersigned also finds that an analysis of whether Defendant has standing or whether the warrant is constitutionally valid is not necessary. Alternatively, in the event the presiding United States District Judge disagrees with the Undersigned's findings and recommendations above and instead finds that the good-faith exception does not apply, the Undersigned recommends that the presiding District Judge recommit the matter to the Undersigned for further analysis on the parties' other arguments.
CONCLUSION
Accordingly, the Undersigned RESPECTFULLY RECOMMENDS that:
1. Defendant's Motion to Suppress Evidence From a Geofence Warrant and Request for Evidentiary Hearing (Doc. 93) be DENIED.
2. Alternatively, in the event the presiding United States District Judge disagrees with the Undersigned's findings and recommendations above and instead finds that the good-faith exception does not apply, the Undersigned recommends that the presiding District Judge recommit the matter to the Undersigned for further analysis on the parties' other arguments.
NOTICE TO PARTIES
A party has fourteen days from the date the party is served a copy of this Report and Recommendation to file written objections to the Report and Recommendation's factual findings and legal conclusions. 28 U.S.C. § 636(b)(1)(C). A party's failure to file written objections waives that party's right to challenge on appeal any unobjected-to factual finding or legal conclusion the district judge adopts from the Report and Recommendation. See 11th Cir. R. 3-1.