Willie James Clements, Jr. v. Commonwealth of Virginia

Opinion

COURT OF APPEALS OF VIRGINIA Record No. 2032-24-4

WILLIE JAMES CLEMENTS, JR.

v.

COMMONWEALTH OF VIRGINIA

Present: Judges O’Brien, Chaney and Callins Argued at Alexandria, Virginia Opinion Issued August 11, 2026

FROM THE CIRCUIT COURT OF ARLINGTON COUNTY Daniel S. Fiore, II, Judge

Kelsey Bulger, Deputy Appellate Counsel (Virginia Indigent Defense Commission, on briefs), for appellant.

Nassir Aboreden, Deputy Commonwealth’s Attorney (Parisa Dehghani-Tafti, Commonwealth’s Attorney; Abhimanyu Mehta, Deputy Commonwealth’s Attorney, on brief), for appellee.

Amici Curiae: Electronic Frontier Foundation, American Civil Liberties Union, and American Civil Liberties Union of Virginia (Matthew William Callahan; Andrew Gellis Crocker, on brief), for appellant.

PUBLISHED OPINION BY

JUDGE DOMINIQUE A. CALLINS

Perhaps it reduces to a matter about a smartphone. But in an era where the average American walks around with the “privacies of life” in their pocket, Riley v. California, 573 U.S. 373, 403 (2014) (quoting Boyd v. United States, 116 U.S. 616, 630 (1886)), this case raises significant questions about the balance between public safety and the constitutional right to privacy, including the extent to which law enforcement may leverage the former against the latter. As the U.S. Supreme Court has noted, “[t]he fact that technology now allows an individual to carry such information in his hand does not make the information any less worthy of the protection for which the Founders fought.” Id. “[T]he government’s unrestrained power

2

to assemble data” from nothing more than the keystrokes a citizen makes on their smartphone charts new territory in the possibilities of law enforcement investigations. United States v. Jones, 565 U.S. 400, 416 (2012) (Sotomayor, J., concurring).

That these questions loom over the jurisprudential horizon is not, however, determinative of the controversy presented here. Willie James Clements, Jr., appeals his conviction for leaving the scene of an accident and aggravated malicious wounding. Arlington police officers obtained a reverse-keyword search warrant to compel Google to identify Clements’s Google account and, ultimately, Clements himself. Clements argues the trial court erred in failing to suppress the evidence from a keyword-search warrant and in failing to grant his motion to strike the Commonwealth’s evidence as to causation. Moreover, Clements contends that, as an initial matter, the trial judge should have recused himself from the suppression hearing. We hold that the trial judge had no obligation to recuse himself after having signed the search warrant, that the exclusionary rule’s good-faith exception allows admission of evidence from the search, and that the causation finding was not plainly wrong or without evidentiary support. Thus, we affirm the trial court’s judgment.

BACKGROUND1

Around 2:20 a.m. on May 15, 2022, video footage outside an Arlington bar showed M.K.

holding her phone and “stumbling” about after ordering a rideshare to take her home. M.K. later testified that it had “been a long day” and that she “had a couple drinks.” A vehicle matching the description of her rideshare arrived, but a dark SUV with no front tag pulled up behind it.

1 We review the evidence in the light most favorable to the prevailing party in the trial court—here, the Commonwealth. Fary v. Commonwealth, 77 Va. App. 331, 341 (2023) (en banc); Cantrell v. Commonwealth, 65 Va. App. 53, 56 (2015). Further, we unseal facts found in the sealed record only to the extent we must discuss them. Brown v. Va. State Bar ex rel. Sixth Dist. Comm., 302 Va. 234, 240 n.2 (2023). And we refer to the survivor by initials to protect her privacy. Poole v. Commonwealth, 73 Va. App. 357, 360 n.1 (2021).

3

Instead of getting into the rideshare,2 M.K. tried opening the rear door of the dark SUV and eventually got into the front passenger seat. As the SUV was driving around fifty miles per hour, an eyewitness traveling near the car saw “something [get] thrown out of the car.” When they slowed near the object, they realized it was a woman, and surveillance footage from the street confirmed a person had fallen from the moving car. The SUV never returned to the scene.

When the police began searching for the SUV driver, they obtained M.K.’s cell phone location data, because they suspected her missing cell phone remained in the car after the suspect pushed her out. Historical cell-site location data led to a neighborhood in Maryland, where police found a “Black Chevrolet Traverse, [with] no front tag,” matching the description of the suspect’s vehicle. Police identified Clements as the suspect after finding the vehicle registered in his name.

M.K. had been pushed out of the SUV only half a mile from her home, and M.K.’s cellphone data “indicate[d] that the phone traveled . . . past [her] residence” before crossing the intersection where the incident occurred. With no evidence indicating the suspect knew M.K. and based on her appearing “extremely intoxicated” in the bar surveillance footage, police reasoned that in order to drive past M.K.’s home, the suspect would have searched her address in a “navigation system” to get directions to her home. On this premise, the police secured a keyword-search warrant3 for any Google account that searched M.K.’s address in Google the night of the incident. Specifically, the warrant authorized the collection of any accounts that on

2 According to the rideshare records, M.K.’s ride was canceled, and no pickup occurred.

3 “Reverse-keyword warrants operate differently than traditional warrants. With traditional warrants, investigators first identify a suspect or suspects, then obtain a warrant to search them or their property for evidence.” People v. Seymour, 536 P.3d 1260, 1268 (Colo. 2023). “However, reverse-keyword warrants start with a potentially incriminating piece of evidence—a search term like the address where the alleged [crime] occurred—then request a list of users implicated by that evidence.” Id.

4

May 15, 2022, from 2:15 a.m. to 3:15 a.m., searched M.K.’s address, or “any reasonable derivatives.” The warrant resulted in only one Google account that searched for her address at 2:22 a.m. on that date. After securing a second search warrant for the account’s information and the phone number associated with it, the police confirmed that the account was associated with Clements.4 In addition to M.K.’s address, Clements’s search history also included, “Arlington VA News, Police, Woman Jumps out of Vehicle,” and “Any suspect in Arlington VA News, Police, Woman Jumps out of Vehicle.”

A grand jury indicted Clements for leaving the scene of an accident and aggravated malicious wounding. Prior to trial, Clements moved to suppress the evidence discovered through the keyword-search warrant, arguing he had a reasonable expectation of privacy over what he searched. Clements also moved to recuse the trial judge from the case because he was the same judge who granted and signed the keyword-search warrant. In his motion, Clements compared the trial judge presiding over both proceedings to a “motion to reconsider,” noting that, unlike such a motion “where the court listens to an issue again with additional information,” his “motion to suppress brings into question the original ruling” and asks the court to “come to a different conclusion than [it] did previously.” The judge denied Clements’s motions, declining to recuse himself and finding probable cause supported the keyword-search warrant.

Clements moved to strike at the close of the Commonwealth’s case and again at the close of all evidence, arguing that the evidence was insufficient to prove causation, but the court denied both motions. The jury convicted Clements of leaving the scene of an accident resulting in injury to a person and aggravated malicious wounding. The trial court denied Clements’s motion to set aside the jury verdict. This appeal followed.

4 Because it was associated with his account as a “recovery email,” Clements’s sister’s email address was also identified in the search. However, the record does not indicate that the sister’s email account itself was searched.

5

ANALYSIS

Clements argues that the trial court erred in denying his motions. He contends that (1) the trial judge should have recused himself, (2) the trial court should have suppressed the evidence discovered from the keyword-search warrant, and (3) the trial court should have granted his motion to strike for causation. We disagree in toto for the reasons that follow.

I. Recusal Clements argues that the trial judge should have recused himself from adjudicating the suppression motion since the trial judge issued the search warrant that yielded the evidence Clements sought to suppress. On this question of first impression, Clements is mistaken.

We review the trial court’s recusal decisions for an abuse of discretion. Prieto v.

Commonwealth, 283 Va. 149, 163 (2012). The trial court abuses its discretion by failing to consider proper factors, giving significant weight to improper factors, or committing a clear error of judgment. Lawlor v. Commonwealth, 285 Va. 187, 213 (2013). Put differently, we will reverse the trial court’s recusal decision only if no reasonable jurist could agree with the trial judge’s ruling. Commonwealth v. Holland, 304 Va. 34, 47 (2025).

Clements bears the burden to prove the trial judge’s bias or prejudice and that it required recusal. Commonwealth v. Jackson, 267 Va. 226, 229 (2004). Due process requires recusal if the trial judge has “‘a direct, personal, substantial, [or] pecuniary interest’ in the outcome of a case.” Welsh v. Commonwealth, 14 Va. App. 300, 314 (1992) (quoting Ward v. Village of Monroeville, 409 U.S. 57, 60 (1972)). Barring those palpably improper interests, we measure whether the trial judge should recuse by “whether he or she harbors ‘such bias or prejudice as would deny the defendant a fair trial.’” Id. at 315 (quoting Justus v. Commonwealth, 222 Va. 667, 673 (1981)). In doing so, we are mindful that a judge should ordinarily recuse when a litigant may question the judge’s impartiality. See generally Va. Sup. Ct. R. pt. 6, § III, Canon 1.

6

But the Canons of Judicial Conduct do not control whether recusal is required. Wilson v. Commonwealth, 272 Va. 19, 28 (2006). “A purported violation of the Canons alone is not enough to mandate [recusal].” Jackson, 267 Va. at 229.

Our Supreme Court has consistently held that a trial judge’s familiarity with a party and their legal issues through prior hearings “does not automatically” raise a question of bias. Deahl v. Winchester Dep’t of Soc. Servs., 224 Va. 664, 672-73 (1983) (quoting Barry v. Sigler, 373 F.2d 835, 836 (8th Cir. 1967)). Nor does a judge possessing—and even expressing—an opinion on the merits of a case, alone, render the judge biased or prejudiced. See Justus, 222 Va. at 673 (“We do not feel that a per se rule requiring a judge to recuse himself on retrial of a capital murder case is necessary to assure a fair trial.”); see also Slayton v. Commonwealth, 185 Va. 371, 376 (1946) (holding that “neither the forming nor the expression” of a factual conclusion disqualifies a judge in a subsequent proceeding); Mason v. Commonwealth, 219 Va. 1091, 1097 (1979) (“[A] judge is not disqualified to sit in a criminal case when, in the disposition of a matter arising out of the same facts, he has formed or expressed an opinion as to the guilt of the accused.”). Moreover, a judge’s prior adverse ruling is not a basis for recusal. Stamper v. Commonwealth, 228 Va. 707, 714 (1985).5 As our Supreme Court noted some time ago, if an adverse ruling “were the criterion of prejudice, no rulings could ever be made which a party opposes.” Id.

Clements’s bald assertion that a trial judge cannot impartially review their own decisions is untethered from Virginia law. Virginia trial judges are regularly compelled to review their own decisions. Renewed motions to strike, motions to set aside a verdict, and motions for reconsideration—all part of regular pleading practice—each ask the same judge who originally

5 Superseded by statute on other grounds, 2021 Va. Acts Sp. Sess. I chs. 523, 540 (codified at Code § 19.2-271.6), as recognized in Shaw v. Commonwealth, 304 Va. 217 (2025).

7

ruled in the same proceeding to re-evaluate their decision based on new arguments or evidence. See, e.g., Williams v. Panter, 83 Va. App. 520, 548 (2025) (“A motion to reconsider ordinarily asks a court to reconsider a holding because, in the opinion of the movant, the holding was erroneous.” (quoting Wal-Mart Stores E., LP v. State Corp. Comm’n, 299 Va. 57, 76 (2020))). And at trial, judges review evidentiary objections taken under advisement or reasserted after being preliminarily overruled based on the development of evidence over the course of the case, notwithstanding a lawyer making substantially the same legal argument as at first blush.

Here, we hold that the trial court did not abuse its discretion by denying Clements’s motion to recuse. The search warrant, issued by the judge in a prior proceeding, reflected only the information contained within the warrant application and affidavit, and did not imbue the judge with extrajudicial knowledge of disputed facts. Nor does the record show actual bias against Clements or in favor of the Commonwealth or otherwise establish an objectively reasonable basis to question the trial judge’s impartiality. And a judge harbors no inherent bias by hearing a motion to suppress evidence seized under a warrant authorized by the same judge in a separate proceeding.

We agree with the trial judge that it was “wholly proper” for him to preside over the subsequent suppression hearing. Confronted with information challenging the validity of his ruling in the ancillary warrant proceeding, the judge—like any judge so confronted—reevaluated the prior ruling. Such reevaluation of a probable cause determination occurred no differently than had the argument been presented before a different judge. And contrary to Clements’s framing, on a motion to suppress, the same or a different judge does not merely “look at the exact same facts, circumstances, and law, but come to a different conclusion.” Instead, the judge is presented with additional facts and circumstances (and, perhaps, law) to consider: those presented by the movant. That the additional information is considered by the same judge, alone,

8

is insufficient to demonstrate partiality. The trial judge appropriately distinguished search warrant hearings from suppression hearings, noting that the former are “determined ex parte,” whereas the latter feature an adversarial proceeding in which a judge may be “persuaded either way.” In so doing, the judge demonstrated his impartiality and lack of bias, and that he understood the appropriate standard governing the suppression hearing.6 Even if the trial judge’s suppression ruling was tainted by an appearance of bias, such appearance of bias must have “affected the outcome of the case” to justify reversal. Welsh, 14 Va. App. at 317. Clements fails to convince us that the alleged appearance of bias affected the outcome of his case, and thus we find his argument unavailing.

II. Keyword-Search Warrant Next, Clements challenges the trial court’s denial of his motion to suppress the keyword-search evidence. He argues that the keyword-search warrant constituted a general warrant barred under the United States and Virginia constitutions. And he avers that the warrant lacked particularity and was unsupported by probable cause. We disagree that the evidence should have been suppressed, holding that the good-faith exception applies to preclude exclusion.

We review de novo the trial court’s denial of Clements’s motion to suppress. McArthur v. Commonwealth, 72 Va. App. 352, 359 (2020). But in doing so, we are bound by the trial court’s factual findings unless plainly wrong or without evidentiary support. Id.

6 Clements argues that the trial judge’s characterization of his role at the suppression hearing as a “reconsideration” demonstrated his bias. Yet, as noted above, Clements introduced the comparison. Further, by affirming the trial court, we are not endorsing this characterization, but merely rejecting Clements’s contention that a trial judge cannot engage in any degree of reconsideration of a suppression ruling without compromising their ethical duty of impartiality.

9

The Fourth Amendment to the United States Constitution7 bars “unreasonable searches and seizures.” This prohibition includes proscribing general warrants. Harvey v. Commonwealth, 76 Va. App. 436, 463 (2023). General warrants are those warrants authorizing searches of “suspected places without evidence of a fact committed, or to seize any person or persons not named, or whose offense is not particularly described and supported by evidence.” Morton v. Commonwealth, 16 Va. App. 946, 952 (1993) (Benton, J., dissenting) (quoting Va. Const. art. I, § 10). Ordinarily, the exclusionary rule bars introduction of evidence seized in violation of the prohibition against general warrants. See Mapp v. Ohio, 367 U.S. 643, 655 (1961) (“We hold that all evidence obtained by searches and seizures in violation of the Constitution is, by that same authority, inadmissible in a state court.”). But since the exclusionary rule is meant to deter police misconduct, Adams v. Commonwealth, 48 Va. App. 737, 746 (2006), we determine whether suppression is required on a case-by-case basis, Midkiff v. Commonwealth, 54 Va. App. 323, 330 (2009). See Janis v. Commonwealth, 22 Va. App. 646, 653 (1996) (“[T]he exclusionary rule is designed to deter police misconduct rather than to punish the errors of judges and magistrates.” (alteration in original) (quoting United States v. Leon, 468 U.S. 897, 916 (1984))).

Among the Fourth Amendment’s many exceptions, the good-faith exception prevents evidentiary suppression in certain circumstances. Leon, 468 U.S. at 920-22. The exclusionary

7 Our Court has previously opined that “protections afforded under the Virginia Constitution are co-extensive with those in the United States Constitution.” Bennefield v. Commonwealth, 21 Va. App. 729, 739-40 (1996). But our Supreme Court has recently acknowledged that we are “absolutely free to interpret state constitutional provisions to accord greater protection” than analogous provisions of the United States Constitution. Vlaming v. W. Point Sch. Bd., 302 Va. 504, 528 (2023) (quoting Arizona v. Evans, 514 U.S. 1, 8 (1995)). Clements appears to capitalize on this acknowledgment, urging us to follow the lead of Colorado on the issue, whose Supreme Court concluded that its constitutional search and seizure provisions extended more protections than the federal ones. Seymour, 536 P.3d at 1272. Because we decide the question presented today on good-faith grounds, we do not opine on whether our search and seizure provisions are similarly more expansive than the federal ones.

10

rule’s deterrent effect loses its benefits when law enforcement officers act in objective good faith by conducting a search within the scope of a warrant obtained from a magistrate. Polston v. Commonwealth, 255 Va. 500, 503 (1998). In those circumstances, we will not suppress evidence except:

(1) Where the magistrate was misled by information in the affidavit which the affiant knew was false or should have known was false, (2) the issuing magistrate totally abandoned his judicial role, (3) the warrant was based on an affidavit “so lacking in indicia of probable cause” as to render official belief in its existence unreasonable or (4) where the warrant was so facially deficient that an executing officer could not reasonably have assumed it was valid.

Colaw v. Commonwealth, 32 Va. App. 806, 811 (2000) (quoting Atkins v. Commonwealth, 9 Va. App. 462, 464 (1990)).

Clements argues the latter two circumstances apply here to override the good-faith exception to suppression. For the former, a supporting affidavit is “so lacking in indicia of probable cause” when it fails to assert “some nexus between the evidence sought and the place to be searched.” Cunningham v. Commonwealth, 49 Va. App. 605, 619 (2007). The Fourth Amendment’s warrant requirement demands more than a “bare bones” affidavit supported by conclusory allegations. Colaw, 32 Va. App. at 813. But if there is “some indicia of probable cause in the underlying affidavit, we will apply the good[-]faith exception.” Anzualda v. Commonwealth, 44 Va. App. 764, 781 (2005). For the facial-deficiency argument, the good-faith exception offers no refuge for warrants that fail to “particularize the place to be searched or the things to be seized.” Leon, 468 U.S. at 923.

Here, the warrant’s supporting affidavit was not “so lacking in indicia of probable cause”

such that suppression was necessary. Colaw, 32 Va. App. at 811 (quoting Atkins, 9 Va. App. at 464). As laid out in their 22-page affidavit, the investigating officers knew that M.K. did not know the suspect, and so the suspect would not have known her address. Since M.K. was

11

“extremely intoxicated,” police knew she likely would not have been able to provide the suspect with directions. As a result, the police could reasonably conclude that the suspect would have had to search M.K.’s address to have driven past her home. And absent evidence that M.K. used her own cellphone to provide navigation, the magistrate could reasonably infer that the suspect used a search application on their phone to navigate to M.K.’s home. See Gwinn v. Commonwealth, 16 Va. App. 972, 975 (1993) (“A magistrate is entitled to draw reasonable inferences about where incriminating evidence is likely to be found.”). Based on all these facts, the law enforcement officers could support a “fair probability that contraband or evidence of a crime w[ould] be found” in the results from Google. United States v. Grubbs, 547 U.S. 90, 95 (2006). The fact that the officers could not have known whether Clements possessed a phone at all is of no moment since probable cause “does not demand any showing that” the probability of finding evidence of a crime would “be correct or more likely true than false.” Curley v. Commonwealth, 295 Va. 616, 622 (2018) (quoting Evans v. Commonwealth, 290 Va. 277, 287 (2015)). This probability only capitalized when the initial search of Google’s records returned only a single account.

Neither was the warrant itself so “facially deficient” that an officer should have known it was not valid. The record reflects the officers’ reasonable efforts to legally obtain information regarding the suspect’s search history. Their detailed affidavit made plain the likelihood that they would discover evidence relating to their criminal investigation through this search. Ornelas v. United States, 517 U.S. 690, 696 (1996) (defining probable cause as the quantum of proof necessary to justify a person’s belief that “contraband or evidence of a crime will be found”). In the absence of controlling guidance on the validity of reverse-keyword searches and rather than shooting in the dark, the officers adapted the three-step warrant procedure discussed

12

by some Virginia circuit courts for geofencing8 warrants.9 Indeed, the officers had no controlling authority addressing the validity of these reverse keyword-search warrants or the procedures to be followed when executing them.

This particular reverse-keyword search warrant is distinguishable from the geofencing warrant considered in Chatrie v. United States, 609 U.S. ___ (2026). There, law enforcement officers obtained a single warrant for data from Google, attempting to cut the judicial system out of the investigatory process thereafter. Id. at ___ (Jackson, J., concurring). They were allowed to access information beyond the scope of the geofence, and they were subject to no duty to narrow their search upon receiving anonymized data from Google. Id. at ___. That same lack of judicial oversight does not exist here. The law enforcement officers in this case narrowly tailored their search to the specific facts of the case and identified precise and limited information. And, importantly, they sought the court’s authorization at each step by obtaining a new warrant in light of the information received from Google in the first instance. This multi-step approach dispels the concerns expressed in Chatrie about supplying a Sovereign with unfettered “access [to] all of a cell-phone user’s movements” such that she would wield a “virtual panopticon with which to scrutinize [] citizens’ activities.” Id. at ___.

8 “Geo-fencing is a process whereby, based on [GPS or] radio frequency identification, the location of an electronic device may be determined.” In re Geo-Fence and Cell Site Location Information Search Warrants, Record No. CM22000505-01, slip op. at 1 n.1 (Arlington Cir. Ct. July 28, 2022) (Fiore, J.)

9 See, e.g., In re Search of Information Stored at the Premises Controlled by Google, Record No. KM-2022-79, slip op. at 10-12 (Fairfax Cir. Ct. Feb. 24, 2022) (Oblon, J.) (analyzing a three-step process for geofencing warrants but requiring police seek the approval of the circuit court before reviewing data received and unmasking particular cell phone numbers); In re Geo-Fence and Cell Site Location Information Search Warrants, slip op. at 6-10 (analyzing the same three-step process for geofencing warrants but deeming the proposed warrant overbroad).

13

We recognize the gravity of this constitutional question, but we must emphasize that we express no view today on the constitutional validity of reverse-keyword search warrants.10 That question may ultimately warrant resolution in an appropriate case. Although the concurrence offers a thoughtful discussion of the issue, there is no dispute that the good-faith exception resolves this appeal without requiring us to reach the underlying constitutional question. Addressing the constitutional question under these circumstances would abandon our obligation to exercise judicial restraint by deciding cases “on the best and narrowest grounds” and thus avoiding the “unnecessary adjudication of a constitutional issue.” Commonwealth v. Swann, 290 Va. 194, 196 (2015) (first quoting McGhee v. Commonwealth, 280 Va. 620, 626 n.4 (2010); and then quoting Bell v. Commonwealth, 264 Va. 172, 203 (2002)); see also Siler v. Louisville & N. R. Co., 213 U.S. 175, 193 (1909) (“Where a case in this court can be decided without reference to questions arising under the Federal Constitution, that course is usually pursued and is not departed from without important reasons.”). This conclusion applies a fortiori to constitutional questions since “[n]o questions can be brought before a judicial tribunal of greater delicacy than those which involve” constitutionality. See McKeithen v. City of Richmond, 302 Va. 422, 435 n.2 (2023) (alteration in original) (quoting Ex parte Randolph, 20 F. Cas. 242, 254 (C.C.D. Va. 1833)). So, although jurists are entitled to voice any concern in a separate writing, principles of

10 We also recognize that the warrants in this case are supported by significantly more indicia of probable cause and are more narrowly tailored in scope than those in Chatrie. And as the U.S. Supreme Court has now clarified that law enforcement’s attempts to obtain detailed digital information from third-parties may constitute searches, we need not provide further guidance to law enforcement officers and magistrates on this point. See Leon, 468 U.S. at 925.

14

judicial restraint and constitutional avoidance still operate to limit our discussions to the issues actually before the Court.11 III. Sufficiency Clements concludes his appeal by arguing that the evidence failed to prove that he caused M.K.’s injury. Again, we disagree.

When reviewing the sufficiency of the evidence, we will not reverse the trial court’s judgment unless plainly wrong or without evidentiary support. Commonwealth v. Moseley, 293 Va. 455, 463 (2017). The touchstone of our inquiry is whether “any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt.” Id. (emphasis added) (quoting Williams v. Commonwealth, 278 Va. 190, 193 (2009)).

Here, the jury’s finding that Clements caused M.K.’s injury is not plainly wrong or without evidentiary support. Under Code § 18.2-51.2(A), the Commonwealth must prove that Clements “cause[d] bodily injury, with the intent to maim, disfigure, disable or kill” M.K. by “shoot[ing], stab[bing], cut[ting] or wound[ing]” her. The jury heard testimony and viewed video evidence showing the driver of the vehicle pushing M.K. from the car, from which she fell headfirst onto the concrete. M.K. had injuries to the back of her head and scuffs on the sides of her shoes, consistent with Clements pushing her from the moving car. Taken together with evidence that Clements searched for M.K.’s address just prior to the incident, a reasonable factfinder could conclude that Clements pushed M.K. out of the car, causing her injuries.

11 Because the constitutional question is unnecessary to the disposition of this appeal, we decline to substantively comment on or endorse the concurrence. Cf. Vlaming, 302 Va. at 584 n.44 (“[A] [concurring] opinion is generally not the best source of legal advice on how to comply with the majority opinion.” (quoting Students for Fair Admissions, Inc. v. President & Fellows of Harv. Coll., 600 U.S. 181, 230 (2023))).

15

CONCLUSION

For the foregoing reasons, we affirm the trial court’s judgment.

Affirmed.

16

Chaney, J., concurring in judgment only.

I concur in the judgment affirming Clements’s convictions and respectfully write separately.

First, I write separately to clarify why the circuit court did not abuse its discretion in denying Clements’s motion to recuse. Although the majority expressly declines to endorse Clements’s characterization of the suppression hearing as a “reconsideration,” it nevertheless describes the judge as having “reevaluated” his earlier probable cause ruling. Maj. Op. at 7, 8 n.6. I would not frame the recusal issue in those terms. Using the term “reevaluated” risks implying that the suppression hearing afforded the issuing judge an opportunity to defend or reaffirm his prior warrant ruling. Instead, I would affirm the circuit court’s denial of recusal on the narrower ground that the issuing judge’s prior involvement was limited to a judicial probable-cause determination. This limited determination did not give the judge personal or extrajudicial knowledge of disputed facts or create an objectively reasonable basis to question the judge’s impartiality.

Second, I write separately to address an issue of first impression in Virginia—whether a reverse-keyword warrant12 directing Google to search its user-query records for unknown persons who searched a specific address during a specific time window satisfies the Fourth Amendment’s probable cause and particularity requirements. Since reverse-keyword warrants present a recurring constitutional question in the digital-search context, it is important to address the merits.

I would conclude that the reverse-keyword warrant was constitutionally defective because the affidavit did not establish the nexus required to search Google’s records specifically,

12 A “reverse-keyword warrant” is also referred to as a “reverse-keyword search warrant,” “keyword search warrant,” or “reverse search warrant.” For consistency, this separate opinion uses “reverse-keyword” warrant throughout.

17

and the warrant authorized a reverse search of unknown users’ query data without particularized probable cause as to any identified person, account, device, or IP address. A Fourth Amendment violation, however, does not necessarily require suppression. On this record, the officers sought and obtained judicial authorization in 2022, before any controlling authority had addressed reverse-keyword warrants, and the affidavit—though constitutionally insufficient—was not so lacking in factual content that their reliance was objectively unreasonable.

For these reasons, I concur in the judgment only.

I. RECUSAL

The circuit court did not abuse its discretion in denying Clements’s motion to recuse.

Although the majority disclaims endorsing Clements’s characterization of the suppression hearing as a “reconsideration,” it continues to describe the judge as having “reevaluated” his prior ruling. Maj. Op. at 7. I would not rely on that framing or an analogy to a motion for reconsideration. The relevant question is not whether a judge may revisit a prior legal ruling in the abstract, but whether this judge’s prior involvement created actual bias, personal or extrajudicial knowledge of disputed facts, or an objectively reasonable basis to question impartiality.

On this record, it did not. The judge’s prior role in issuing the warrant did not, standing alone, give him personal or extrajudicial knowledge of the disputed facts. Rather, his prior involvement reflected judicial knowledge obtained from reviewing the warrant application and affidavit. The judge expressly grounded his ruling on the distinction between “personal knowledge” and “judicial knowledge.” R. 849. He explained that, unlike a judge who issues a wiretap order—who must recuse because the intercepted communication is filed with the court and may become a source of personal knowledge, see Code § 19.2-68—a judge who issues an ordinary search warrant “is only making a legal determination,” and is “not testing th[e] facts.”

18

R. 873-74. The record also reflects that the judge had multiple warrant applications before him and signed some while declining others, including a related geofence warrant he refused to sign. R. 848. A judge who grants one application and denies another exercises the judgment of a neutral magistrate; that role does not constitute grounds for recusal.

I do not, however, rest on the broader suggestion that recusal can never be required simply because the General Assembly has mandated it for wiretap issuing judges but not for those who issue ordinary search warrants. Our Supreme Court has held that recusal may be warranted even in the absence of a statutory command. See Wilson v. Commonwealth, 272 Va. 19, 30 (2006). On the narrower point, on this record, the judge’s purely legal warrant determination did not provide the personal or extrajudicial knowledge of disputed facts that recusal requires.

Nor does this record show actual bias, favoritism toward the Commonwealth, or an objectively reasonable basis to question the judge’s impartiality. At the recusal hearing, the judge described the warrant’s search term as “extremely narrow in its focus.” R. 869. However, under established Virginia law, neither forming nor expressing an opinion on an issue in a prior proceeding disqualifies a judge from ruling on that matter later. See Slayton v. Commonwealth, 185 Va. 371, 376 (1946); Mason v. Commonwealth, 219 Va. 1091, 1098 (1979). An initial remark made while explaining why recusal was unnecessary does not establish the bias or extrajudicial knowledge that would mandate recusal.

Although the judge described the motion as asking the court “basically to reconsider its prior legal determination,” the record shows that he understood the distinction between the ex parte warrant proceeding and the later suppression hearing. R. 869. The judge’s reference to “reconsideration” is best understood as a response to Clements’s suggestion that a judge may never revisit a prior ruling, not as a description of the governing suppression standard. The judge

19

invoked the routine practice of reconsidering rulings to explain why he was not categorically disqualified from hearing the motion. He also recognized that the suppression hearing differed from the ex parte warrant proceeding because it allowed him to “be persuaded either way.” R. 851-52, 870-71.

The judge further explained that the warrant proceeding had been ex parte, while the suppression hearing allowed the court to consider additional adversarial briefing and argument. R. 870-71. In ruling on the suppression motion, the court did not merely reaffirm its earlier probable-cause determination. It considered the parties’ arguments, applied the suppression standard and legal framework, and ruled against the Commonwealth in part by concluding that Clements had a reasonable expectation of privacy in his Google search data. R. 497.

Thus, I would not adopt any broad rule suggesting that suppression hearings are properly understood as reconsideration of earlier warrant decisions. Nor would I suggest that a judge’s prior issuance of a warrant is categorically irrelevant to a later recusal motion. The narrower point is sufficient. This judge’s limited prior judicial involvement did not create personal or extrajudicial knowledge of disputed facts or otherwise require recusal. I, therefore, concur in the judgment affirming the denial of Clements’s recusal motion.

II. FOURTH AMENDMENT GOVERNING FRAMEWORK The Fourth Amendment to the United States Constitution protects against unreasonable searches and seizures. “The ‘basic purpose of this Amendment,’ . . . ‘is to safeguard the privacy and security of individuals against arbitrary invasions by governmental officials.’” Carpenter v. United States, 585 U.S. 296, 303 (2018) (quoting Camara v. Mun. Ct. of City & Cnty. of San Francisco, 387 U.S. 523, 528 (1967)). “The Founding generation crafted the Fourth Amendment as a ‘response to the reviled “general warrants” and “writs of assistance” of the colonial era,

20

which allowed British officers to rummage through homes in an unrestrained search for evidence of criminal activity.’” Id. (quoting Riley v. California, 573 U.S. 373, 403 (2014)).

The Warrant Clause provides in relevant part, “no Warrants shall issue, but upon probable cause, supported by Oath or affirmation, and particularly describing the place to be searched, and the persons or things to be seized.” U.S. Const. amend. IV. The Supreme Court of the United States has determined the words of the Warrant Clause of the Fourth Amendment “are precise and clear.” Stanford v. Texas, 379 U.S. 476, 481 (1965). “They reflect the determination of those who wrote the Bill of Rights that the people of this new Nation should forever ‘be secure in their persons, houses, papers, and effects’ from intrusion and seizure by officers acting under the unbridled authority of a general warrant.” Id. (emphasis added) (quoting U.S. Const. amend. IV).

Article I, Section 10 of the Virginia Constitution reflects the same historical concern, providing that general warrants authorizing officers “to search suspected places without evidence of a fact committed,” or to seize persons “not named” or whose offense is “not particularly described and supported by evidence,” are “grievous and oppressive, and ought not to be granted.” Va. Const. art. I, § 10. “Our courts have consistently held that the protections afforded under the Virginia Constitution are co-extensive with those in the United States Constitution.”13 Patrick v. Commonwealth, 50 Va. App. 650, 654 n.4 (2007) (quoting Henry v. Commonwealth, 32 Va. App. 547, 551 (2000)). “[T]he ultimate touchstone of the Fourth Amendment is

13 Although this Court has described the protections afforded by the Virginia Constitution as co-extensive with those afforded by the U.S. Constitution, Virginia courts remain “absolutely free to interpret state constitutional provisions to accord greater protection” than analogous federal provisions. Vlaming v. W. Point Sch. Bd., 302 Va. 504, 528 (2023) (quoting Arizona v. Evans, 514 U.S. 1, 8 (1995)). Since I would conclude that the warrant was defective under the Fourth Amendment, I do not decide whether Article I, Section 10 affords greater protection in this context.

21

‘reasonableness.’” Jones v. Commonwealth, 71 Va. App. 375, 380-81 (2019) (quoting Kentucky v. King, 563 U.S. 452, 459 (2011)).

The particularity requirement exists to “limit the discretion that police officers may exercise when executing a search warrant and to preclude them from engaging in a fishing expedition or an ‘exploratory rummaging’ in a person’s belongings.” Morke v. Commonwealth, 14 Va. App. 496, 500 (1992) (quoting Coolidge v. New Hampshire, 403 U.S. 443, 467 (1971)); see also Moyer v. Commonwealth, 33 Va. App. 8, 24 (2000) (“However, search warrants that limit the executing officers’ discretion by directing them to seize only evidence of a specific crime consistently have been held to satisfy the particularity requirement of the [F]ourth [A]mendment.” (alterations in original) (quoting Morke, 14 Va. App. at 500-01)).

Our test for assessing the particularity required by the Fourth Amendment is pragmatic.

This Court recognizes that “[t]he degree of specificity required . . . may necessarily vary according to the circumstances and type of items involved.” Morke, 14 Va. App. at 500 (citation omitted) (quoting United States v. Torch, 609 F.2d 1088, 1090 (4th Cir. 1979), cert. denied, 446 U.S. 957 (1980)). “[T]here is a practical margin of flexibility permitted . . . in the description of items to be seized.” Id. (alterations in original). “So long as the ‘search warrant describes the objects of the search with reasonable specificity,’ it complies with the dictates of the [F]ourth [A]mendment.”14 Id. (quoting United States v. Reed, 726 F.2d 339, 342 (7th Cir. 1984)).

14 Cf. Jeffers v. Commonwealth, 62 Va. App. 151, 158 (2013) (“Here, the search warrant identified various items ‘tending to . . . give evidence . . . of child pornography,’ as the ‘specific things to be searched for and seized,’ and named ‘106 Barricks Mill Rd.,’ as the ‘property to which entry is sought.’ Thus, the scope of the search was ‘defined by the object of the search and the places in which there is probable cause to believe that it may be found.’ Police had traced child pornography to this address, but did ‘not know in which of the [buildings] the illegal transactions [were] taking place.’” (alterations in original) (first quoting Zurcher v. Stanford Daily, 436 U.S. 547, 556 (1978); then quoting Maryland v. Garrison, 480 U.S. 79, 84 (1987); and then quoting id. at 88 n.13 (noting that this situation presents “quite different issues” from the case where it becomes evident “that a valid warrant describes too broadly the premises to be searched”))).

22

“[T]he existence of probable cause is determined by examining the totality-of-the-

circumstances.” Taylor v. Commonwealth, 66 Va. App. 619, 635 (2016) (alteration in original) (quoting Anzualda v. Commonwealth, 44 Va. App. 764, 774 (2005)). “Probable cause for the issuance of a search warrant exists when, ‘given all the circumstances. . . , there is a fair probability that contraband or evidence of a crime will be found in a particular place.’” Id. (alteration in original) (quoting Tart v. Commonwealth, 17 Va. App. 384, 387 (1993)). Thus, in line with the particularity requirement, probable cause requires a nexus between the evidence sought and the place to be searched. See Cunningham v. Commonwealth, 49 Va. App. 605, 619 (2007) (explaining that an affidavit may be so lacking in probable cause where it fails to assert “some nexus between the evidence sought and the place to be searched”).

The particularity and probable cause principles do not lose force because the search occurs in the digital world. To the contrary, “modern cell phones . . . implicate privacy concerns far beyond those implicated by the search of a cigarette pack, a wallet, or a purse.” Riley, 573 U.S. at 393. The Supreme Court has also recognized that digital devices can contain the “privacies of life.” Id. at 403 (quoting Boyd v. United States, 116 U.S. 616, 630 (1886)). As the majority recognizes, “[t]he fact that technology now allows an individual to carry such information in his hand does not make the information any less worthy of the protection for which the Founders fought.” Id.; see also Maj. Op. at 1.

The United States Supreme Court’s recent decision in Chatrie v. United States, 609 U.S.

___ (2026), confirms, in the context of Google Location History, that neither a limited period nor third-party possession necessarily defeats Fourth Amendment protection. The Court held that police conducted a search when they obtained two hours of Google Location History, explaining: “It does not matter if the time period scrutinized was only two hours. Nor does it matter that the materials obtained were handed over by a third-party tech company.” Id. at ___. Chatrie

23

concerned cell-phone location information, not search-query records, and thus does not itself resolve the distinct question presented but reinforces that familiar Fourth Amendment principles continue to govern novel forms of digital information.15 Id. at ___.

Those principles inform the analysis here: whether the affidavit established a nexus to Google’s records and whether the warrant confined the authorized search with sufficient particularity.

III. ANALYSIS

This appeal requires applying familiar Fourth Amendment requirements to a novel digital-search tool—a reverse-keyword warrant that begins with search terms rather than with a known suspect, account, device, or IP address. Clements argues that “the warrant failed to identify any particular suspect or the specific data that it sought to search.” Op. Br. 22. He also argues that the warrant lacked a factual nexus between the alleged illegal activity and “Google user data.” Op. Br. 25. Amici describe reverse-keyword warrants as compelling Google to search “its entire reserve of user data” for a search term specified by police. Am. Br. 3. The principal constitutional concern raised by Clements and amici is that the warrant authorized the government to work backward from a search term to a person, rather than to search a particularized person, account, device, or place for evidence supported by probable cause.

The Commonwealth emphasizes that the warrant was limited to a single address, a one-

hour window, and the time period during which M.K. was in the vehicle. A’ee Br. 33-34. I

15 Chatrie also rejected application of the third-party doctrine to Google Location History. 609 U.S. at ___. The Supreme Court explained that the information was “not truly shared” because “[t]he exposure of that information to Google is merely what happens when a user avails himself of one of the services on his cell phone.” Id. at ___. The Court therefore concluded: “So just as the third-party doctrine did not apply in Carpenter, it does not apply here.” Id. at ___.

Although Chatrie concerned Location History rather than search-query records, its analysis bears on whether a user relinquishes Fourth Amendment protection merely because sensitive digital information is maintained by a technology company.

24

recognize the importance of those limits, particularly to the good-faith inquiry. However, those facts do not eliminate the Fourth Amendment issue. The affidavit’s specific facts supported an inference that the driver may have used navigation, but they did not establish probable cause to search Google search-query data for unknown users to discover who may have searched the address.

As an initial matter, I agree with the majority that the good-faith exception provides the narrowest ground for the Court’s disposition of this appeal. That does not make a separate discussion of the preserved constitutional question improper. Chatrie recently reiterated Leon’s instruction that courts may “resolv[e] the Fourth Amendment issue” before the good-faith question either to better assess good faith or “to guide future action by law enforcement officers and magistrates.” Chatrie, 609 U.S. at ____ n.4 (alteration in original) (quoting United States v. Leon, 468 U.S. 897, 925 (1984)).

The distinction between the probable cause and good-faith inquiries matters. Although the majority states that it expresses “no view” on the warrant’s constitutional validity, its good- faith analysis reasons that the police “could reasonably conclude” that the suspect searched the address, that the magistrate “could reasonably infer” that the suspect used a search application, and that the facts supported a “fair probability” that evidence would be found in Google’s results. Maj. Op. at 11, 13. Those conclusions bear directly on the probable cause nexus, not merely on whether the affidavit contained “some indicia of probable cause” sufficient to permit objectively reasonable reliance.

I, therefore, explain why I disagree with the majority’s nexus analysis while agreeing that the affidavit was not so lacking in factual content that the good-faith exception fails.

25

A. The reverse-keyword warrant is constitutionally defective.16 The reverse-keyword warrant authorized the following search:

This warrant applies to information associated with the following search term (or reasonably derived terms) as used during the time frame of May 15, 2022, at the time of 0215 hours to 0315 hours (EDT) (UTC-4)

801 North Florida Street Arlington, Virginia 801 N. Florida Street Arlington, Virginia 801 North Florida St Arlington, Virginia 801 N. Florida St Arlington, Virginia

And any reasonable derivatives of these terms.

that are stored at premises owned, maintained, controlled, or operated by Google, a company headquartered in Mountain View, California.

R. 2190 (emphases in original).

For any responsive search, the warrant required Google to disclose:

(a) All IP addresses associated with the search terms (b) Date and time of the search (c) All User Agent Strings associated with IPs executing the searches matching the search terms (d) Correlation between the IP address(es) and any known Google accounts, including but not limited to Google Email, Google Hangouts, Google Drive, Android, Youtube, and Google Chat.

16 Chatrie does not resolve whether this initial warrant satisfied the Fourth Amendment’s probable-cause and particularity requirements. The Supreme Court held that obtaining Google Location History constituted a search, but expressly left to the United States Court of Appeals for the Fourth Circuit whether each step of the geofence process was supported by probable cause and described with particularity. Chatrie, 609 U.S. at ___. The Court did not answer the different defect presented here by the initial warrant: the absence of a factual nexus to Google and of particularized probable cause as to an identified person, account, device, or IP address.

Justice Ketanji Brown Jackson separately concluded that “at a minimum,” the second and third stages of the geofence process were defective because officers could obtain additional information without meaningful narrowing criteria or renewed review by a magistrate. Chatrie, 609 U.S. at ___ (Jackson, J., concurring). The officers here obtained additional judicial authorization before seeking further account information, which distinguishes this case from Chatrie’s particular later-stage concern identified in Justice Jackson’s concurrence.

26

(e) Account information for accounts that correlated to the aforementioned search terms to include any accounts linked via “html cookies.”17

(f) Correlation between the IP address(es) and any known devices, including but not limited to Android, iOS, etc.

R. 2191. Based on its plain language, the warrant was not directed at a known suspect, account, device, or IP address. Instead, it required Google to identify accounts that searched for one of four variants of M.K.’s address, or “any reasonable derivatives” of those terms, during the one- hour period from 2:15 a.m. to 3:15 a.m. on May 15, 2022. R. 2190. The affidavit described the thing to be searched as “Google LLC records pertaining to the Google Accounts that searched a particular search term during specific times.” R. 2168. Thus, when the warrant was issued, law enforcement had not identified the person, account, device, or IP address to be searched.

The warrant instead used the search term to identify the suspect. The affidavit supplied reason to believe that the driver may have used some form of navigation. R. 2178. It stated that M.K. appeared “extremely intoxicated,” that the suspect vehicle drove past her home, and that there was no evidence the suspect knew her. R. 2178. From those facts, the affiant concluded that the suspect “utilized some type of navigation system” to locate the victim’s residence. R. 2178.

That conclusion may be a reasonable investigative inference. Even assuming the circumstances supported an inference that the driver used a cellphone or a digital navigation application, the affidavit supplied no fact indicating that the driver used Google rather than Apple Maps or another navigation service. R. 2178; Op. Br. 26. The affidavit stated that Google

17 As Electronic Frontier Foundation, American Civil Liberties Union, and American Civil Liberties Union of Virginia, amici, explained, some publicly reported warrants have been used to obtain IP addresses, account information, and “User Agent Strings,” web browsing tracking information called “cookies” or “similar information that can reveal identity.” Am. Br. 13 (citing Google Decl. 7). According to amici, because keyword warrants require Google to search user data for specified terms, they may implicate innocent people who happen to search for something potentially incriminating.

27

possesses a navigation system accessible through Google Maps or a general Google search. R. 2188. However, that general description of Google’s capabilities did not connect this driver or this offense to Google. Nor could Google’s general popularity or functionality, standing alone, supply the missing nexus. The widespread availability of a service does not establish a fair probability that evidence of this offense would be found in that service’s records. The affidavit also did not rely on the direct-route trajectory of the suspect vehicle as evidence of Google navigation use. R. 2178.

Detective Bamford’s suppression hearing testimony confirms the existence of this gap.

When questioned about the basis for the warrant’s scope, he stated that the one-hour timeframe was determined by the timing of the offense—specifically, the victim’s cell phone remaining in the vehicle, a witness’s observation of the suspect vehicle turning off Florida Street, and the time the victim was thrown—but identified no factual connection to Google. R. 892. Thus, although the temporal parameters were plausibly justified, the linkage to Google was absent. The warrant therefore moved from the inference that the driver likely used a navigation system to a command that Google search its user-query records, without establishing the factual bridge that probable cause requires between the suspected activity and the specific place to be searched.

That distinction matters because probable cause requires a “fair probability” that evidence will be found in the particular place to be searched. Taylor, 66 Va. App. at 635 (quoting Tart, 17 Va. App. at 387). The place to be searched here was Google’s records. R. 2168, 2190. A warrant directed to Google’s records, therefore, required some factual basis to believe that evidence of the offense would be found there, not merely that the driver used some form of navigation. Cunningham v. Commonwealth, 49 Va. App. at 617, illustrates the point. There, this Court explained that the generalized proposition that drug users may sometimes keep contraband in their homes does not establish the nexus needed to search a particular defendant’s

28

residence. Id. at 616-17. The principle is the same here: the general inference that a driver in these circumstances likely used some navigation tool does not, by itself, establish that evidence will be found in Google’s records specifically. Although Cunningham involved a residence and not a digital service provider, its doctrinal principle emphasizes that a generalized inference cannot replace the required nexus between the evidence sought and the place to be searched.

The warrant also did not satisfy the particularity requirement merely because it was limited to one hour and variants of one address. Those limits are relevant and narrow the scope of the warrant. However, on these facts, I would assess particularity by reference to what the warrant authorized Google to search at the outset, not by what Google ultimately returned. This warrant did not identify a known suspect, account, device, or IP address. Instead, it directed Google to search its records of unknown users’ query data to determine who, if anyone, had searched for the victim’s address during that period. R. 2168, 2190. That structure is materially different from a warrant that identifies a known person, account, device, or place to be searched and then limits the evidence to be seized.

Although the reverse-keyword search produced a single Google account, that does not resolve the constitutional inquiry. The warrant’s correlation provisions also caused Clements’s sister’s email address to be identified because it was linked to Clements’s account as a recovery email, although the record does not show that her account itself was searched. R. 893-94, 2191; Maj. Op. at 4 n.4. That disclosure illustrates that the warrant’s correlation provisions could reveal identifying information concerning a person who was not identified as having searched the address, based solely on her incidental connection to the responsive account.

The Fourth Amendment protects against the search itself, not the amount of information ultimately provided or produced. Here, the warrant’s correlation provisions revealed identifying information associated with a person who had not been identified as having searched the address,

29

solely because her email address was linked to Clements’s account as a recovery email. The warrant’s reference to “any reasonable derivatives” of the listed search terms reinforces that the constitutional right protects against the search itself. R. 2190. The particularity requirement exists to “limit the discretion that police officers may exercise when executing a search warrant.” Morke, 14 Va. App. at 500. Neither the warrant nor the affidavit defines what constitutes a “reasonable derivative” of the listed address variants. That leaves Google and law enforcement with discretion to determine the outer bounds of the search. I do not suggest this alone would be dispositive in every case, but on this record, it supports the conclusion that the warrant was not sufficiently particularized.

A search may be unconstitutionally broad at the outset, even if the result of the search identifies only one person. The fact that the search ultimately yielded only one responsive account cannot retroactively supply the missing nexus. In rejecting a results-based approach to determining whether surveillance constituted a search, Chatrie explained that “[w]hether something is a search does not depend on what it finds.” 609 U.S. at ___; see also United States v. Di Re, 332 U.S. 581, 595 (1948) (“[A] search is not to be made legal by what it turns up. In law it is good or bad when it starts.”). The warrant’s validity here likewise must be assessed from the information available when it is issued, not from the evidence the search ultimately produced.

Persuasive authority in the context of geofence warrants underscores this point. In United States v. Smith, 110 F.4th 817, 837 (5th Cir. 2024), the United States Court of Appeals for the Fifth Circuit addressed a different digital-search tool but identified a similar reverse search concern: “While the results of a geofence warrant may be narrowly tailored, the search itself is not.” Although a geofence is materially different from a reverse-keyword warrant, the

30

same reverse search concern is present here.18 Carpenter v. United States, 585 U.S. 296, also suggests that courts should focus on the constitutionality of the search rather than the amount of information ultimately returned. In Carpenter, the Supreme Court’s analysis accounted for the government’s ability to use the potential digital data to reconstruct private information that historically would have been difficult to obtain. See 585 U.S. at 311-13. The same principle matters here. The warrant required a provider to search otherwise difficult-to-obtain data associated with unknown users before law enforcement had particularized probable cause as to any user, account, device, or IP address. R. 2191.

The constitutional concern is not only the volume of information that may be searched, but also the character of the information. Search queries can reveal what a person seeks to know, where a person may be going, what a person believes or fears.19 Am. Br. 15-16. Amici in this case observe that reverse-keyword warrants may implicate interests beyond ordinary identifying

18 In Chatrie, the Supreme Court of the United States defined geofence warrants and described the process, stating:

In recent years, law enforcement officers have employed so-called geofence warrants to obtain information that technology companies collect about their users’ cell-phone locations. Suppose that investigators know a crime was committed at a particular place and time, but do not have a suspect. They may draw a “geofence”—a virtual perimeter—around the crime scene and get a warrant compelling a company to hand over data about the cell phones located in that area near the time of the crime. Following a process specified in the warrant, the company will turn over the cell-phone data and eventually identify by name one or more of the users thus disclosed.

609 U.S. at ___.

19 Chatrie remarked upon the invasion of an individual’s privacy when officers accessed private data, by stating, “‘[e]ven short-term monitoring’ of a person’s physical movements can provide ‘a wealth of detail about [a person’s] familial, political, professional, religious, and sexual associations.’” Chatrie, 609 U.S. at ___ (quoting United States v. Jones, 565 U.S. 400, 415 (2012) (Sotomayor, J., concurring)).

31

information because they reach into thought, inquiry, and belief. Am. Br. 3, 15-17, 20, 27. The trial court itself questioned whether searches for medical, political, or religious information carry an expectation of privacy. R. 903. Although this is not a First Amendment case, that expressive dimension reinforces the need to apply the probable cause and particularity requirements with care. See Stanford, 379 U.S. at 485 (requiring “the most scrupulous exactitude” when a search implicates expressive materials).

I, therefore, would hold that this warrant was constitutionally defective. This does not mean that all reverse-keyword warrants are categorically invalid. I would hold only that, on this record, the warrant lacked particularized probable cause connecting Google user data to the suspected offender and authorized a search that worked backward from a search term to a suspect.20 Such a holding would not leave law enforcement without a path forward. The Commonwealth identified that path itself. It acknowledged that, before seeking the keyword warrant, police had already connected the suspect vehicle to Clements through the victim’s phone data and that the available next step would have been a “search warrant for his Google account, just his.” R. 914-15. A warrant directed at an identified account is the ordinary, particularized course; the constitutional defect here arose only because the warrant instead worked backward from a search term through the data of unknown users. A different affidavit might supply that nexus through evidence that the suspect used a smartphone with Google Maps

20 Carpenter v. United States, 585 U.S. 296, is instructive. There, the Supreme Court of the United States held that the government’s acquisition of historical cell-site location information was a search under the Fourth Amendment and explained that the digital data at issue did “not fit neatly under existing precedents.” Id. at 306, 309-10. Although Carpenter involved cell-site location information rather than search-query data, and although this case involves a warrant, Carpenter instructs courts to be cautious in extending older doctrines to new digital tools that permit the government to obtain information in ways that were not previously possible. Id. at 310-11. Applying that caution here, I would hold only that this warrant was defective because it lacked particularized probable cause connecting Google user-query data to the suspected offender and instead worked backward from a search term to a suspect.

32

or another Google navigation service, that GPS or Google Maps data was recovered from a relevant device or account, or that digital-forensic evidence linked a device associated with the suspected offender to Google’s infrastructure through location-history data, Google account sign-ins, or device-correlation information. These examples are not exhaustive, and I express no view on whether any particular showing would be sufficient in a different case.

It is enough here to say that the affidavit needed some fact linking the suspected use of navigation specifically to Google’s records. Since the affidavit supplied no fact connecting this driver or this offense to Google, it did not establish the required nexus.

B. The officers’ reliance on the warrant was objectively reasonable.

Although the warrant was constitutionally defective, that conclusion does not require suppression. See Washington v. Commonwealth, 60 Va. App. 427, 435 (2012) (“[T]he ‘fact that a Fourth Amendment violation occurred—i.e., that a search or arrest was unreasonable—does not necessarily mean that the exclusionary rule applies.’” (quoting Herring v. United States, 555 U.S. 135, 140 (2009))). “[T]he exclusionary rule is designed to deter police misconduct rather than to punish the errors of judges and magistrates.” Janis v. Commonwealth, 22 Va. App. 646, 653 (1996) (alteration in original) (quoting Leon, 468 U.S. at 916). Thus, “suppression of evidence obtained pursuant to a warrant should be ordered only on a case-by-case basis and only in those unusual cases in which exclusion will further the purposes of the exclusionary rule.” Polston v. Commonwealth, 255 Va. 500, 503 (1998) (quoting Leon, 468 U.S. at 918).

Under the good-faith exception, evidence obtained pursuant to a warrant is not suppressed unless, among other circumstances, the affidavit was “so lacking in indicia of probable cause” that official belief in its validity was unreasonable, or the warrant was “so facially deficient” that an executing officer could not reasonably presume it valid. Colaw v. Commonwealth, 32 Va. App. 806, 811 (2000) (quoting Atkins v. Commonwealth, 9

33

Va. App. 462, 464 (1990)). The “pertinent analysis of deterrence and culpability is objective” and “confined to the objectively ascertainable question whether a reasonably well[]trained officer would have known that the search was illegal” in light of “all of the circumstances.” Collins v. Commonwealth, 297 Va. 207, 219 (2019) (citing Herring, 555 U.S. at 145). “An officer’s decision to obtain a warrant is prima facie evidence that he or she was acting in good faith.” Adams v. Commonwealth, 275 Va. 260, 273 (2008) (quoting United States v. Koerth, 312 F.3d 862, 868 (7th Cir. 2002)).

Since the exclusionary rule is aimed at deterrence, “evidence obtained from a search should be suppressed only if it can be said that the law enforcement officer had knowledge, or may properly be charged with knowledge, that the search was unconstitutional under the Fourth Amendment.” Ward v. Commonwealth, 273 Va. 211, 224-25 (2007) (quoting Leon, 468 U.S. at 919-21). Thus, when an officer’s conduct is objectively reasonable, exclusion does not further the purposes of the rule “in any appreciable way” because it “in no way affect[s an officer’s] future conduct[.]” Leon, 468 U.S. at 920 (quoting Stone v. Powell, 428 U.S. 465, 539-40 (1976) (White, J., dissenting)). “This is particularly true . . . when an officer acting with objective good faith has obtained a search warrant from a judge or magistrate and acted within its scope.” Id.; see also Bellamy v. Commonwealth, 60 Va. App. 125, 133 (2012) (explaining that when an officer does not commit a “systemic error or [exhibit] reckless disregard of constitutional requirements,” the exclusionary rule’s deterrence effect is not furthered (quoting Herring, 555 U.S. at 147-48)).

Chatrie reinforces the constitutional significance of government access to sensitive digital information, but it does not resolve the exclusionary rule issue presented here. The Supreme Court expressly left open both whether the geofence warrant satisfied the Fourth Amendment’s probable cause and particularity requirements and whether the good-faith

34

exception permits admission of the evidence obtained. 609 U.S. at ___, ___ & n.4 (leaving probable cause and particularity for remand); id. at ___ n.4 (declining to decide good faith). More importantly, the good-faith inquiry is objective and must account for the legal landscape when the officers acted. Thus, although Chatrie strengthens the conclusion that the initial warrant was constitutionally defective, it does not establish that the officers’ reliance on a judicially authorized warrant in 2022 was objectively unreasonable.

Two circumstances are especially relevant to the objective-reliance inquiry here: the officers sought and obtained a judicially authorized warrant before obtaining the records, and they acted in 2022, before any controlling authority had addressed reverse-keyword warrants. See Parady v. Commonwealth, 78 Va. App. 18, 38 (2023) (“In examining ‘the state of the law at the time of the search,’ we ask whether a ‘“reasonably well trained officer would have known that the search was illegal” in light of “all of the circumstances.”’” (quoting Collins, 297 Va. at 215)); Adams, 275 Va. at 273 (“An officer’s decision to obtain a warrant is prima facie evidence that he or she [is] acting in good faith.” (quoting Koerth, 312 F.3d at 868)).

Clements argues that the officers’ decision to seek a warrant shows they understood constitutional protections were implicated and that this awareness should weigh against good faith. However, the good-faith exception is designed, in part, to encourage officers to obtain judicial authorization before acting rather than to proceed without a warrant. See Leon, 468 U.S. at 920 (“This is particularly true . . . when an officer acting with objective good faith has obtained a search warrant from a judge or magistrate and acted within its scope.”). Treating warrant-seeking as evidence of bad faith would undermine that purpose. The key question is not whether the officers recognized a constitutional issue, but whether their reliance on the warrant issued was objectively reasonable.

35

In this context, the good-faith inquiry asks whether a reasonable officer could rely on the warrant issued, not whether the officer correctly anticipated every constitutional limitation in a novel unsettled area of digital search law. See Adams, 275 Va. at 275 (“An officer . . . is not required to go behind a magistrate’s probable cause determination to ascertain whether probable cause actually existed.”); cf. Reed v. Commonwealth, 71 Va. App. 164, 174 (2019) (explaining that “[u]nless a statute is clearly unconstitutional, an officer cannot be expected to question the judgment of the legislature that passed the law,” and if a “statute is subsequently declared unconstitutional, excluding evidence obtained pursuant to it prior to such a judicial declaration will not deter future Fourth Amendment violations” (quoting Illinois v. Krull, 480 U.S. 340, 349-50 (1987))).

On this record, both circumstances support the conclusion that the officers’ reliance on the judicially authorized warrant was objectively reasonable. I therefore cannot conclude that a reasonably well-trained officer would have known that reliance on the warrant was unlawful. A warrant may be constitutionally defective without being so lacking in indicia of probable cause that no reasonable officer could rely on it. Those are different thresholds. One indication that reliance was objectively reasonable is whether the affidavit “provided evidence sufficient to create disagreement among thoughtful and competent judges as to the existence of probable cause.” Tart, 17 Va. App. at 390 (citing Leon, 468 U.S. at 926).

Probable cause requires a fair probability that evidence will be found in the place to be searched; the good-faith inquiry asks whether the affidavit was so deficient—so bare of any factual predicate—that official reliance was unreasonable. See Smith v. Commonwealth, 56 Va. App. 592, 600 (2010) (“In dealing with probable cause, . . . as the very name implies, we deal with probabilities. These are not technical; they are the factual and practical considerations of everyday life on which reasonable and prudent men, not legal technicians, act.” (alteration in

36

original) (quoting Brinegar v. United States, 338 U.S. 160, 176 (1949))); Colaw, 32 Va. App. at 811; Anzualda, 44 Va. App. at 774.

The officers sought judicial authorization before executing the search. They submitted an affidavit describing the investigation, identified a specific address, limited the request to a one- hour window supported by the victim’s cell phone location data and tracking of the suspect vehicle, and obtained a warrant based on a reasoned—if ultimately constitutionally insufficient— chain of inference from the facts of the offense. R. 2171-86, 2190. Although the affidavit failed to establish the Google-specific nexus required for probable cause, it was not so bare of factual content that no reasonable officer could rely on the warrant. See, e.g., Smith, 110 F.4th at 838- 39; People v. Seymour, 536 P.3d 1260, 1278 (Colo. 2023). Chatrie did not establish a contrary good-faith rule. See 609 U.S. at ___ n.4.

Nor was the warrant so facially deficient that an executing officer could not reasonably presume it valid. See Colaw, 32 Va. App. at 811. As explained above, the warrant’s “reasonable derivatives” language is undefined and raises a particularity concern. R. 2190. However, facial deficiency in the good-faith context concerns whether a warrant is so lacking on its face that an officer executing it cannot identify what it authorizes.

Here, the warrant identified the provider, the specific search terms, the one-hour time period, and the categories of data to be produced. R. 2190-91. A reasonable officer reading the face of the warrant could have understood “reasonable derivatives” as extending to ordinary variations of the listed address terms, such as abbreviations or alternate spellings of the street name or street type, consistent with the four variants the warrant itself listed. That interpretation may not address the issue of constitutional particularity, but it is sufficient to ensure this warrant does not fall into the limited category of facially deficient warrants excluded from the good-faith exception.

37

The state of the law when the officers acted supports the applicability of the good-faith exception. When this warrant was issued, no controlling authority had addressed the constitutional requirements governing reverse-keyword warrants. Later decisions involving related digital search tools illustrate the novelty and continuing difficulty of the issue, but they could not have informed the officers’ conduct in 2022. See Smith, 110 F.4th at 838 (holding a geofence warrant constitutes a Fourth Amendment search but declining to suppress evidence under the good-faith exception); Seymour, 536 P.3d at 1278 (assuming without deciding a defect in probable cause; applying good-faith exception to a reverse-keyword warrant). Given the absence of controlling authority at the time, the warrant’s temporal and search-term limits, and the officers’ decision to obtain judicial authorization, the officers’ reliance was not objectively unreasonable.

IV. CONCLUSION

For these reasons, I would affirm the denial of Clements’s recusal motion on narrower grounds than those expressed by the majority. I would hold that, on this record, the Google reverse-keyword warrant violated the Fourth Amendment because the affidavit did not establish the nexus required to search Google’s records specifically, and the warrant authorized a reverse search of unknown users’ query data without particularized probable cause as to any identified person, account, device, or IP address.

However, when officers sought the warrant and executed the search in 2022, no controlling authority had addressed reverse-keyword warrants; the officers sought and obtained judicial authorization; and their reliance on the warrant was not objectively unreasonable. Chatrie does not alter this conclusion because it neither decided the validity of a reverse- keyword search warrant nor resolved the application of the good-faith exception. I, therefore, concur in the judgment affirming Clements’s convictions.