STATE v. GASPER (2026)

¶1 This is a review of a published court of appeals decision, State v. Gasper, 2024 WI App 72, 414 Wis. 2d 532, 16 N.W.3d 279, reversing the Waukesha County circuit court's order granting defendant Michael Joseph Gasper's motion to suppress evidence.

¶2 Gasper was charged with ten counts of possessing child pornography ¹ and nine counts of child exploitation based upon the content on his cell phone.² Law enforcement obtained a warrant for his cell phone after the National Center for Missing and Exploited Children (“NCMEC”) forwarded a CyberTipline report ³ (which included a single, flagged, 16-second video) from Snapchat to the Wisconsin Department of Justice (“DOJ”). No person at Snapchat or NCMEC viewed the contents. Instead, Snapchat scanned its platform and identified the video file it flagged as known CSAM using a hash-based scanning program. The flagged video was first viewed by a person when an employee of the DOJ did so without a warrant. Then the CyberTip with the flagged video was forwarded to local law enforcement who also viewed the video without obtaining a warrant. Gasper seeks to suppress this evidence on the basis that it was obtained in violation of his Fourth Amendment rights.

¶3 The circuit court granted Gasper's motion to suppress all evidence of CSAM on the basis that there was a warrantless search of his cell phone which violated the Fourth Amendment to the United States Constitution under Riley v. California, 573 U.S. 373, 134 S.Ct. 2473, 189 L.Ed.2d 430 (2014), and Carpenter v. United States, 585 U.S. 296, 138 S.Ct. 2206, 201 L.Ed.2d 507 (2018). The circuit court also determined that suppression was appropriate because Message-Digest 5 (“MD5”),⁴ a traditional hash-value scanning program, is “not secure,” stating “[c]ollision verification is clearly important in the private party search doctrine. With MD5 specifically at issue in Gasper's case, it should not be relied upon.”

¶4 The court of appeals reversed in a published opinion determining that there was no Fourth Amendment violation because Gasper did not have a reasonable expectation of privacy. Gasper, 414 Wis. 2d 532, ¶¶15-16, 16 N.W.3d 279. The court of appeals concluded that even if Gasper had a subjective expectation of privacy, his “obviously unlawful” conduct violated Snapchat's terms of service and any subjective expectation that he had was “objectively unreasonable given Snapchat's policies regarding sexual content in general and sexually explicit content involving children in particular.” Id., ¶22. The court of appeals concluded that no Fourth Amendment search occurred. Id., ¶¶1, 29.

¶5 The Fourth Amendment serves as a limit on government power. See Camara v. Mun. Ct. of City & Cnty. of San Francisco, 387 U.S. 523, 528, 87 S.Ct. 1727, 18 L.Ed.2d 930 (1967). (“The basic purpose of this Amendment, as recognized in countless decisions of this Court, is to safeguard the privacy and security of individuals against arbitrary invasions by government officials.”); Carpenter, 585 U.S. at 305, 138 S.Ct. 2206 (explaining that the Fourth Amendment places hurdles “in the way of a too permeating police surveillance”). A private search is not a government search. United States v. Ginglen, 467 F.3d 1071, 1074 (7th Cir. 2006). The Fourth Amendment is inapplicable to a search which has been completed by a private party as that search frustrates an individual's expectation of privacy. United States v. Jacobsen, 466 U.S. 109, 117, 104 S.Ct. 1652, 80 L.Ed.2d 85 (1984). The Fourth Amendment is implicated, however, if the government exceeds the private search. Id. at 115–22, 104 S.Ct. 1652. Gasper does not argue that the government viewed more than the one video provided, nor does he argue that anything else of significance was in the video. Gasper relies entirely on the argument that the government exceeded Snapchat's private search because a person in the government was the first to open and view the video, and did so without a warrant.

¶6 We conclude that the private search doctrine applies. It is undisputed that Snapchat performed a private search ⁵ when it scanned and flagged the single, 16-second video as CSAM. The government did not exceed the scope of Snapchat's search when it viewed the video because any expectation of privacy Gasper may have had in the video was frustrated by the private search, and there was virtual certainty that law enforcement would not find anything of significance beyond what the private search revealed. As a result, the Fourth Amendment is not implicated. Accordingly, we affirm the court of appeals and remand to the circuit court for further proceedings consistent with this opinion.⁶

I. FACTUAL AND PROCEDURAL BACKGROUND

¶7 On January 13, 2023, Snapchat reported that it found CSAM and submitted a single, flagged, 16-second video to NCMEC. Snapchat's hash-based scanning program, Microsoft's PhotoDNA,⁷ had detected and flagged a CSAM video that had been uploaded to Snapchat's servers from Gasper's account. PhotoDNA “scans files to determine if they are copies of known and reported [CSAM] based on their ‘hash values.’ ”⁸ Gasper, 414 Wis. 2d 532, ¶2, 16 N.W.3d 279. No person at Snapchat viewed the video.

¶8 NCMEC also did not view the video, but did confirm that the video was CSAM through a hash match of the uploaded file to visually similar files that were previously viewed and categorized by NCMEC.⁹ NCMEC sent a CyberTip and the flagged video to the DOJ. A DOJ analyst opened the video file, without a warrant, to confirm that it contained CSAM. After confirming the video file contained CSAM, DOJ issued an administrative subpoena to the internet service provider to obtain the name and mailing address associated with the account.

¶9 The internet service provider responded with the account information which implicated Gasper. DOJ forwarded the CyberTip and attached video to the Waukesha County Sheriff's Office. There, a detective trained in this area opened the video, without a warrant, and also “confirmed that it depicted [CSAM].” Id., ¶4. Based upon the CyberTip and video content, the detective then applied for, received, and executed a search warrant for Gasper's home and electronic devices. Police discovered ten files on Gasper's cell phone containing CSAM. Gasper was taken into custody, waived his Miranda ¹⁰ rights, and admitted that he had accessed and stored CSAM on his cell phone.

¶10 Gasper was charged with ten counts of Possession of Child Pornography, in violation of Wis. Stat. § 948.12(1m) and (3)(a) (2023–24),¹¹ and nine counts of Sexual Exploitation of a Child, in violation of Wis. Stat. § 948.05(1m) and (2p)(a). Gasper moved to suppress the evidence on the basis that it was an unconstitutional search in violation of the Fourth Amendment to the United States Constitution. He argued that because the government was the first to view the video and did so without a warrant, the search was unconstitutional. He also argued that the evidence recovered from the subsequent search warrant is likewise unconstitutional, being fruit of the initial warrantless, unconstitutional search of the Snapchat video.

¶11 The circuit court held a hearing on Gasper's motion to suppress. The Waukesha County Sheriff's Office detective was the only witness. He described how CyberTips and PhotoDNA operate. He said Snapchat uses PhotoDNA. PhotoDNA is “its own thing, its own program. It's its own software.” The detective testified that PhotoDNA is, in his experience, “a reliable source of identifying suspected [CSAM.]” The detective testified that if he receives a CyberTip that has a PhotoDNA match or hash match, “[he has] never had it not be pornographic.” In this case, knowing PhotoDNA was used, he said “it's likely going to be child sexual abuse material.” PhotoDNA was the “sole thing that was used” to detect this CSAM.

¶12 The detective said PhotoDNA's analysis does not use an MD5 hash value for the overall file. He testified that PhotoDNA works by analyzing pieces of a file and comparing the similarity of those pieces to previously identified CSAM. Even if PhotoDNA would have used an MD5-based algorithm, there was no indication that a risk of collision (an incorrect identification or false positive match) would be present. The detective was questioned about the theoretical risk of collision, but he stated that he observed no evidence of it in this case, and that collision had only been observed in laboratory settings with extremely small sized files. NCMEC, however, does provide an MD5 hash value on the CyberTip to assist in the future investigation. The detective referenced the MD5 hash value in his affidavit to the search warrant, but he stated that MD5 was not used in this case.

¶13 The State submitted into evidence Snapchat's user agreement and policies and conditions which specifically “banned [CSAM]” and informed its users that Snapchat was actively scanning for CSAM on its platform. Its user agreement and policies and conditions also informed users that Snapchat's discovery of CSAM will be reported to NCMEC and law enforcement.

¶14 Gasper did not testify, but he attempted to submit an affidavit which detailed the steps he took to keep his Snapchat account, cell phone, and Wi-Fi private and password protected. The circuit court denied his request to submit the affidavit, but accepted his statements regarding his expectation of privacy as an offer of proof.

¶15 The circuit court granted Gasper's motion to suppress stating that “[t]here is a legitimate privacy interest in cell phones.” The court concluded that the private search doctrine is inapplicable for two reasons: (1) no human at Snapchat “eyeballed” the image, and (2) PhotoDNA assigned an MD5 hash value to the video and MD5 is categorically unreliable because of the risk of collision. The circuit court seemed to rely on information found on a website rather than the detective's testimony to conclude that MD5 is subject to “collision” and that the technology used was “not secure.”¹² The circuit court stated,

This case shows why proving probable cause on a case by case basis remains important. Relying on algorithms and computer programs as a basis for avoiding warrants is like relying on the ever changing waters of a river because technology and its limits change so fast. Ultimately, such reliance that expands the existing private/third party doctrine is for higher courts to decide.

¶16 The court of appeals reversed the circuit court's decision and concluded that Gasper did not have a reasonable expectation of privacy:

A search occurs for the purpose of the Fourth Amendment “when an expectation of privacy that society is prepared to consider reasonable is infringed.” State v. Purtell, 2014 WI 101, ¶21, 358 Wis. 2d 212, 851 N.W.2d 417 (quoting [Jacobsen, 466 U.S. at 113, 104 S.Ct. 1652]). ․ The privacy interest is both subjective and objective: a defendant must show he or she subjectively expected privacy in the area or object, and that the expectation is one that society recognizes as reasonable.

Gasper, 414 Wis. 2d 532, ¶10, 16 N.W.3d 279 (internal citation omitted).

¶17 The court of appeals stated the circuit court erred by simply relying on Riley, 573 U.S. 373, 134 S.Ct. 2473, 189 L.Ed.2d 430, and Carpenter, 585 U.S. 296, 138 S.Ct. 2206, because the appropriate “area” of the search was Gasper's Snapchat account, not his cell phone, citing its decision in State v. Bowers, 2023 WI App 4, 405 Wis. 2d 716, 985 N.W.2d 123 (2022). Gasper, 414 Wis. 2d 532, ¶14. The court of appeals explained that in Bowers, it “rejected the State's argument that because Bowers created the account with his county government email address and his employer could access the Dropbox account through the email address, Bowers lacked a reasonable expectation of privacy.” Id. (citing Bowers, 405 Wis. 2d 716, ¶¶22, 42, 985 N.W.2d 123). The court of appeals reasoned that the same analysis applied here because, “Snapchat did not access the video in Gasper's account through his cell phone. Rather, the video was obtained directly from Gasper's Snapchat account.” Id., ¶15. The court of appeals went further to address whether Gasper had a reasonable expectation of privacy in the video in his Snapchat account and concluded that he did not. The court of appeals relied primarily on Snapchat's Terms of Service, Community Guidelines, and Sexual Content Explainer. The court of appeals decided that “Gasper has failed to satisfy his burden to prove either his subjective or an objective expectation of privacy.” Id., ¶20.

¶18 Gasper petitioned our court for review, which we granted.

II. STANDARD OF REVIEW

¶19 “ ‘Our review of an order granting or denying a motion to suppress evidence presents a question of constitutional fact.’ ” State v. Tullberg, 2014 WI 134, ¶27, 359 Wis. 2d 421, 857 N.W.2d 120 (quoting State v. Robinson, 2010 WI 80, ¶22, 327 Wis. 2d 302, 786 N.W.2d 463). We accept the circuit court's factual findings unless they are clearly erroneous, and “independently apply constitutional principles to those facts.” Id.

III. ANALYSIS

A. Fourth Amendment Principles: Private Search Doctrine

¶20 The Fourth Amendment to the United States Constitution provides:

The right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated, and 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.

¶21 “ ‘The touchstone of the Fourth Amendment is reasonableness. The Fourth Amendment does not proscribe all state-initiated searches and seizures; it merely proscribes those which are unreasonable.’ ” Tullberg, 359 Wis. 2d 421, ¶29, 857 N.W.2d 120 (internal citations omitted) (quoting Florida v. Jimeno, 500 U.S. 248, 250, 111 S.Ct. 1801, 114 L.Ed.2d 297 (1991)).¹³ Warrantless searches are presumptively unreasonable. Id., ¶30. Fourth Amendment protections are an important check on government action. “[T]he touchstone of [Fourth] Amendment analysis [is] the question [of] whether a person has a ‘constitutionally protected reasonable expectation of privacy.’ ” Oliver v. United States, 466 U.S. 170, 177, 104 S.Ct. 1735, 80 L.Ed.2d 214 (1984) (quoting Katz v. United States, 389 U.S. 347, 360, 88 S.Ct. 507, 19 L.Ed.2d 576 (1967) (Harlan, J., concurring)).

¶22 The Fourth Amendment provides “rights against the government” not private parties. Hiibel v. Sixth Jud. Dist. Ct. Nev., Humboldt Cnty., 542 U.S. 177, 187, 124 S.Ct. 2451, 159 L.Ed.2d 292 (2004). Stated differently, the Fourth Amendment applies to only “governmental action[.] [I]t is wholly inapplicable ‘to a search or seizure, even an unreasonable one, effected by a private individual ․’ ” Jacobsen, 466 U.S. at 113–14, 104 S.Ct. 1652 (quoting Walter v. United States, 447 U.S. 649, 662, 100 S.Ct. 2395, 65 L.Ed.2d 410 (1980) (Blackmun, J., dissenting)). A government agent may “view[ ] what a private party ha[s] freely made available for his inspection” without offending the Fourth Amendment. Id. at 119–20, 104 S.Ct. 1652 (citing Coolidge v. New Hampshire, 403 U.S. 443, 487–90, 91 S.Ct. 2022, 29 L.Ed.2d 564 (1971); Burdeau v. McDowell, 256 U.S. 465, 475–76, 41 S.Ct. 574, 65 L.Ed. 1048 (1921)). This is because a private actor's earlier search frustrates the owner's legitimate expectation of privacy. Id. at 119, 104 S.Ct. 1652. In other words, under the private search doctrine, any expectation of privacy is lost because the private actor's search abrogates the original expectation.¹⁴

It is well settled that when an individual reveals private information to another, he assumes the risk that his confidant will reveal that information to the authorities, and if that occurs the Fourth Amendment does not prohibit governmental use of that information. Once frustration of the original expectation of privacy occurs, the Fourth Amendment does not prohibit governmental use of the now nonprivate information[.]

Id. at 117, 104 S.Ct. 1652. “[T]he critical inquiry under the Fourth Amendment is whether the authorities obtained information with respect to which the defendant's expectation of privacy has not already been frustrated.” United States v. Runyan, 275 F.3d 449, 461 (5th Cir. 2001).

¶23 A private search, when repeated by the government, does not then become a government search, unless that search exceeds the scope of the private search. Jacobsen, 466 U.S. at 115, 104 S.Ct. 1652. Herein lies the crux of Gasper's argument—that the government exceeded Snapchat's private search when it viewed the video without a warrant. Gasper does not argue that the video itself contained anything other than what was represented—CSAM. He also does not argue that the government searched more than the single, 16-second video Snapchat scanned, flagged, and reported. His argument relies entirely on the fact that no person at Snapchat actually looked at the video, and that because a person in the government was the first to view the video, that viewing exceeded Snapchat's private search. But, since the State has asserted that the search was a private search, it is the defendant who bears the burden of proving that a government search occurred, to a preponderance of the evidence. State v. Payano-Roman, 2006 WI 47, ¶23, 290 Wis. 2d 380, 714 N.W.2d 548.

B. Private Party Search: Unconstitutional Expansion Argument

¶24 Gasper agrees that Snapchat is a private party. Gasper argues that law enforcement unconstitutionally exceeded Snapchat's private search because a person in government, not a person at Snapchat, was first to view the video with human eyes. More specifically, Gasper argues that it was unconstitutional for law enforcement to view the video because it “expanded the scope of the computer data scan contained in the CyberTip from NCMEC” and “expanded the scope of Snapchat's private search.”

¶25 Stated differently, Gasper asserts that all of the CSAM evidence should be suppressed because the search violated the Fourth Amendment.¹⁵ He makes much of the fact that no person at Snapchat viewed the video before forwarding it to law enforcement. But, Gasper fails to meet his burden of proving that this was a government search that exceeded the private search. The Fourth Amendment serves as a limit on government power, not a deterrent to private actors, in this case ESPs who use technology to protect the integrity of their platforms, and in so doing, find CSAM.¹⁶ Snapchat's PhotoDNA detected and flagged Gasper's video as CSAM, and Snapchat reported that video to NCMEC, who then forwarded the CyberTip and video to the DOJ, who then forwarded the same to the Waukesha County Sheriff's Office.

¶26 We begin our analysis of the private search doctrine with Walter v. United States, 447 U.S. 649, 100 S.Ct. 2395, 65 L.Ed.2d 410 (1980), wherein the United States Supreme Court suppressed evidence found when the FBI viewed, without a warrant, film strips that were misdelivered to a company. The company received 12 packages that contained 871 boxes of film. Walter, 447 U.S. at 651, 100 S.Ct. 2395. The company's employees opened one or two of the packages and discovered boxes which depicted “suggestive drawings” and had “explicit descriptions.” Id. at 652, 100 S.Ct. 2395. One employee attempted to view film by holding it to the light, but was unsuccessful. Id. The company turned the boxes over to the FBI who reviewed them over the next two months, using a projector, without a warrant. Id. Walter, a plurality opinion, concluded that the FBI search of the film strips was unreasonable because it “constituted an unreasonable invasion of [the film strip] owner's constitutionally protected interest in privacy.” Id. at 654, 100 S.Ct. 2395. The court concluded that the projection of the films by the FBI was “a significant expansion of the search that had been conducted previously by a private party and therefore must be characterized as a separate search.” Id. at 657, 100 S.Ct. 2395. The court noted that “[p]rior to the Government screening, one could only draw inferences about what was on the films.” Id.

¶27 Walter is distinguishable from the case before us today. Unlike the employee's failed attempt to view a sampling of the film strips, Snapchat's PhotoDNA scanned, opened, and flagged the single, 16-second video as CSAM. The government did not have access to any other materials that may have been scanned. Unlike the “private search” in Walter, where one could only infer what might be on the film, there was a virtual certainty that law enforcement would view nothing else of significance beyond what Snapchat's technology scanned and reported, and that law enforcement viewing the video “would not tell [the government] anything more than [it] already had been told.” Jacobsen, 466 U.S. at 119, 104 S.Ct. 1652. And, unlike turning over all 12 misdelivered and mostly unopened packages which contained 871 boxes of film to the FBI, Snapchat provided only the single, flagged, 16-second video. Unlike the FBI personnel who, without a warrant, took months to review all of the film strips with a projector, the government here viewed the single, 16-second video that Snapchat scanned, flagged as CSAM, and turned over to law enforcement.

¶28 The private search doctrine and whether the government exceeds the scope of the private search was next examined in Jacobsen, 466 U.S. 109, 104 S.Ct. 1652. Jacobsen concluded that the Drug Enforcement Administration (“DEA”) “agent's viewing of what a private party had freely made available for his inspection did not violate the Fourth Amendment” because “there was a virtual certainty that nothing else of significance” was in the package, and the inspection would not have provided “anything more than [what it] already had been told.” Id. at 119, 104 S.Ct. 1652.

¶29 In Jacobsen, Federal Express employees accidentally damaged a package with a forklift and then opened it, pursuant to company policy, to prepare an insurance claim. Id. at 111, 104 S.Ct. 1652. The employees discovered several plastic baggies of white powder. Id. Federal Express called the DEA and agents from the DEA arrived. Id. However, before they arrived, the box had been repackaged. Id. A DEA agent opened the repackaged box to test the white powder, which tested positive for cocaine. Id. at 111–12, 104 S.Ct. 1652.

¶30 When referring to the employees’ private search, the Supreme Court stated, “Whether those invasions were accidental or deliberate, and whether they were reasonable or unreasonable, they did not violate the Fourth Amendment because of their private character.” Id. at 115, 104 S.Ct. 1652. The Court then reasoned that although this is a private search, “additional invasions of ․ privacy by the Government agent must be tested by the degree to which they exceeded the scope of the private search.” Id. It determined that “the Fourth Amendment does not prohibit governmental use of the now nonprivate information.” Id. at 117, 104 S.Ct. 1652. However, the Fourth Amendment is implicated “if the authorities use information with respect to which the expectation of privacy has not already been frustrated.” Id. The Supreme Court concluded that the DEA “agent's viewing of what a private party had freely made available for his inspection did not violate the Fourth Amendment” because “there was a virtual certainty that nothing else of significance was in the package.” Id. at 119–20, 104 S.Ct. 1652. The Court further noted that the government's interest was “substantial,” particularly since it was “virtually certain that the substance tested was in fact contraband.” Id. at 125, 104 S.Ct. 1652.¹⁷

¶31 Since Jacobsen, the private search doctrine has been found to be applicable when there is a “virtual certainty” that the government's search will not reveal anything more than that which the private party represented. See United States v. Lichtenberger, 786 F.3d 478, 488 (6th Cir. 2015); Runyan, 275 F.3d at 463–64; United States v. Ackerman, 831 F.3d 1292, 1305–06 (10th Cir. 2016). Although “virtual certainty” has not been specifically defined, it “implies something less than absolute confidence.” United States v. Rivera-Morales, 961 F.3d 1, 11 (1st Cir. 2020). It is “a common-sense determination into whether there is anything more than a remote or highly unlikely possibility that the officer's actions will uncover something of significance apart from what the private searcher has found and reported.” Id.

¶32 Hash value comparison has been regarded as a scanning mechanism that detects CSAM with “ ‘almost absolute certainty.’ ” United States v. Reddick, 900 F.3d 636, 639 (5th Cir. 2018) (quoting United States v. Larman, 547 F. App'x 475, 477 (5th Cir. 2013) (unpublished)). Jacobsen’s virtual certainty standard is met when the inspection “would not tell [the government] anything more than [it] had been told.” Jacobsen, 466 U.S. at 119, 104 S.Ct. 1652.¹⁸ The detective in this case testified that he had never seen a file identified by PhotoDNA as CSAM to be anything other than CSAM. And here, Gasper has not demonstrated that the government's viewing of the 16-second video would reveal anything of significance beyond that which Snapchat scanned using PhotoDNA. The single, flagged video was removed from Gasper's account and included as part of Snapchat's CyberTip. Stated differently, “the government does not conduct a Fourth Amendment search when there is a ‘virtual certainty’ that its search will disclose nothing more than what a private party's earlier search has revealed.” United States v. Miller, 982 F.3d 412, 417–18 (6th Cir. 2020) (quoting Jacobsen, 466 U.S. at 119, 104 S.Ct. 1652). That is the case here.

¶33 We note that the Sixth Circuit in Miller, 982 F.3d 412, and the Fifth Circuit in Reddick, 900 F.3d 636, under similar facts, concluded that there is no Fourth Amendment violation because the private search doctrine applied. Gasper, however, urges us to instead adopt the reasoning of the Ninth and Second Circuits in United States v. Wilson, 13 F.4th 961 (9th Cir. 2021), and United States v. Maher, 120 F.4th 297 (2d Cir. 2024), which conclude otherwise. We, like several other states, decline Gasper's invitation.¹⁹

¶34 In Miller, a detective viewed two images identified as CSAM by a Google company scan, and the court concluded this was not a Fourth Amendment search under the private search doctrine. The court noted that a hash value is “a sort of digital fingerprint.” Miller, 982 F.3d at 417 (quoting Ackerman, 831 F.3d at 1294). Email files were scanned for certain hash values and matched to a copy of an illegal file. Id. at 420. Google's scan revealed a Gmail account that had uploaded two files with hash values that matched CSAM. Id. Google sent the report with the files and the IP address to NCMEC. Id. NCMEC alerted local law enforcement. Id. Miller argued that the search was unconstitutional—that the police detective conducted an unreasonable search when he opened and viewed the files. Id. at 426. The court, however, relied on the private search doctrine and concluded that the Fourth Amendment restricts government, not private, action. Id. at 417. The court stated that “the government does not conduct a Fourth Amendment search when there is a ‘virtual certainty’ that its search will disclose nothing more than what a private party's earlier search has revealed.” Id. at 417–18 (quoting Jacobsen, 466 U.S. at 119, 104 S.Ct. 1652). The Sixth Circuit concluded that it was Google's technology that “opened” and “inspected” the files, revealing that they had the same content as the known CSAM. Id. at 431. The court determined that “[t]his ․ information satisfies Jacobsen’s virtual-certainty test and triggers its private-search doctrine.” Id. at 430. Relying upon the unchallenged reliability of the hashing technology, the court concluded that the private search doctrine applied because it was virtually certain that the officer's viewing of the files would disclose nothing more than the same images that the private actor's employees had already viewed. Id. at 418.

¶35 The Sixth Circuit held that the government viewing the file did not infringe on a reasonable expectation of privacy or qualify as an unconstitutional search because the conduct did not exceed the scope of the earlier private search. Id. at 430. The court reasoned that it must ask “whether Google's hash-value search of the files using its digital eyes made it virtually certain that [the detective] would discover no more than what Google had learned when he viewed the images with his human eyes.” Id. at 428 (citing Jacobsen, 466 U.S. at 119, 104 S.Ct. 1652). The court concluded that because the detective “viewed only files with hash-value matches[,] ․ the private search doctrine applies.” Id. at 429. The court noted that rather than comparing the detective's viewing of the files to the Jacobsen agent's field test, “we must compare Google's search of the files to the [Federal Express] employees’ search of the box.” Id. The Sixth Circuit concluded that the hash-value match from Google created “virtual certainty” that the investigator would view CSAM upon opening the segregated file. Id. at 417–18, 430. Like Miller, Snapchat's PhotoDNA scanned, flagged, and verified that the files were CSAM.²⁰ “Under the private-search doctrine, the government does not conduct a Fourth Amendment search when there is a ‘virtual certainty’ that its search will disclose nothing more than what a private party's search has revealed.” Id. at 417–18. The detective here was given no more than what Snapchat's scanning technology flagged as CSAM—a single, 16-second video.

¶36 The Miller court went further to explain that a hash-value scanning software may be more reliable than human observation, noting that hash-value scanning software does not contain the same human subjectivity and need for recall. The court referenced the “risk of a flaw in the [person's] recollection,” and noted that if a person has a “quick view” of suspected CSAM, law enforcement would be permitted to conduct a more thorough investigative examination. Id. at 430–31 (quoting Jacobsen, 466 U.S. at 119, 104 S.Ct. 1652) (alteration in original). But if the “view” is conducted by more reliable hash-value scanning, law enforcement would be precluded from having the same ability. Id. at 430. The court understood that unlike the human eye, “[c]ommon hash algorithms, by contrast, catalogue every pixel.” Id. The court questioned, “What sense would it make to treat a more accurate search of a file differently?” Id. at 431. Like in Miller, the record in the case before us demonstrates that Snapchat used PhotoDNA which is reliable in detecting CSAM and does so, unlike a human eye, by performing a pixel-by-pixel analysis. But the private search doctrine does not necessarily depend on the reliability of the information provided. Here, PhotoDNA's hashing function flagged the reported video as it matched its previously scanned, known, CSAM. Even if PhotoDNA was deemed unreliable, which we need not decide, it is less than clear how that would impact the analysis of the private search doctrine since the analysis focuses on what the private actor searched.²¹ The Miller court stated, and we agree, that “[j]ust because a private party turns out to be wrong about the legality of an item that the party discloses to police does not mean that the police violate the Fourth Amendment when they reexamine the item.” Id.

¶37 In Reddick, 900 F.3d 636, the Fifth Circuit Court of Appeals concluded that the detective did not violate Reddick's Fourth Amendment rights by reviewing image files that had been flagged by a cloud hosting service as matching known CSAM files. The court stated:

The private search doctrine decides this case. A private company determined that hash values of files uploaded ․ corresponded to the hash values of known [CSAM] images. The company then passed this information on to law enforcement. This qualifies as a “private search” for Fourth Amendment purposes. And the government's subsequent law enforcement actions in reviewing the images did not effect an intrusion ․ that [Reddick] did not already experience as a result of the private search.

900 F.3d at 637. Reddick argued that the detective's warrantless opening of the files was an unlawful search. Id. at 638. But, “[u]nder the private search doctrine, ‘the critical inquiry under the Fourth Amendment is whether the authorities obtained information with respect to which [Reddick's] expectation of privacy has not already been frustrated.’ ” Id. (quoting Runyan, 275 F.3d at 461). Reddick stated the “hash value comparison ‘allows law enforcement to identify [CSAM] with almost absolute certainty,’ since hash values are ‘specific to the makeup of a particular image's data.’ ” Id. at 639 (quoting Larman, 547 F. App'x at 477). The court concluded that when Reddick uploaded the files, PhotoDNA reviewed the hash values of those files and compared them against known hash values of CSAM. The court stated:

In other words, his “package” (that is, his set of computer files) was inspected and deemed suspicious by a private actor. Accordingly, whatever expectation of privacy Reddick might have had in the hash values of his files was frustrated by Microsoft's private search.

․

[O]pening the file merely confirmed that the flagged file was indeed [CSAM], as suspected.

Id.

¶38 Gasper urges this court to instead adopt the rationale of the Ninth and Second Circuits in Wilson, 13 F.4th 961, and Maher, 120 F.4th 297, which concluded that law enforcement exceeded the scope of the private search. But, a closer look at those cases reveals that their reasoning is at odds with Jacobsen, 466 U.S. 109, 104 S.Ct. 1652, 80 L.Ed.2d 85. Unlike Jacobsen, those courts concluded that expansion of the private search occurs when law enforcement is able to view more precise details about the content of the videos, rather than whether law enforcement was virtually certain to view anything else of significance beyond what the private search revealed. It is incongruous to conclude that a defendant holds no privacy interest in a video flagged as CSAM by a private actor, but simultaneously recognize a privacy interest of CSAM details in the same video.

¶39 Moreover, Wilson and Maher do not seem to even consider how the scanning operates. Before any law enforcement is alerted, an ESP's scan essentially inspects the files and compares them to known CSAM in order to detect the contents. PhotoDNA alerts Snapchat when a file is found to contain CSAM. Snapchat scanned, flagged, and turned over a single, 16-second video, not all files that Snapchat may have scanned. In the case at issue, the detective viewing the reported video could not have expanded Snapchat's private search because in order to detect the CSAM in the first instance, it had already been scanned by Snapchat's PhotoDNA.

¶40 Gasper does not argue that the government search exceeded Snapchat's because it viewed more videos than the one provided or discovered more information from that video. He does not argue that the single, flagged, 16-second video contained anything other than CSAM. Gasper's argument, like in Maher and Wilson, rests solely on the fact that no person at Snapchat viewed the video and because a person in government was the first to view it, its viewing necessarily exceeded the Snapchat search. Gasper's argument assumes that a private search cannot occur unless human eyes view the evidence. He also assumes that when a person views potential CSAM, their assessment, regardless of how long, must be more reliable than the PhotoDNA scanning system. See supra, ¶36. This is not to say that a computer program is infallible, but the record here bears no indication that the PhotoDNA used was somehow flawed. While the private search doctrine most often involves a person who has seen the evidence and then turns it over to law enforcement, the doctrine itself does not require that a person actually view the evidence. Whether a Snapchat employee viewed the video or not is of no moment to the private search doctrine, because the private search doctrine allows the government to review what a private actor has already searched, so long as there is a virtual certainty that its search will disclose nothing more of significance than what the private party search revealed.

¶41 Here, Snapchat scanned, flagged, and reported an “apparent” CSAM video. Of course, law enforcement may examine the exact same video more thoroughly or with a different base of knowledge than a private party so long as there is a virtual certainty that they will not find anything of significance beyond that which the private search revealed. Miller, 982 F.3d at 431; Runyan, 275 F.3d at 464; United States v. Simpson, 904 F.2d 607, 610 (11th Cir. 1990); United States v. Tosti, 733 F.3d 816, 822 (9th Cir. 2013). An officer may “learn” more than a lay person who views the same evidence and may see the details of the CSAM, but that alone does not automatically equate to the government exceeding the private search. The Maher and Wilson courts concluded that by viewing the video, suspected CSAM, law enforcement exceeded the private actor's search because law enforcement saw the details of the CSAM. Those courts reasoned that viewing the video revealed the “particulars” of the CSAM to law enforcement, and in so doing, the private search was exceeded. The private search doctrine is not solely evaluated from the perspective of what details law enforcement might see when viewing the video. Instead, it is evaluated from the perspective of what the private party's search revealed and whether there is a virtual certainty that law enforcement will not find anything else of significance beyond that which the private search revealed.

¶42 Jacobsen did not analyze the Fourth Amendment based upon what law enforcement might see or learn when viewing the same evidence. Jacobsen and its progeny do not depend on the experience and knowledge of the viewer. Whether law enforcement may glean something more from viewing the same file does not amount to an expansion of the initial search. The private search is not exceeded if the identical information, here a single, 16-second video, is scanned and flagged as CSAM by a private actor, then provided to law enforcement for review and they look at the video. The test remains whether the government's search exceeds what the private party's search revealed. Both Wilson and Maher misapply Jacobsen because they turn on the fact that the investigating officers learned the CSAM details. In Jacobsen, the employees suspected the substance was cocaine, yet law enforcement could reopen the packages and test the powder without offending the private search doctrine. Reopening the package to view its contents was not unconstitutional. Learning whether the powder was cocaine was also not prohibited. We do not agree with Maher’s and Wilson’s conclusion that law enforcement exceeds the private actor search when it views the “particulars” in the already-scanned CSAM video. We disagree that seeing the details of the CSAM is an expansion of the private search, see Wilson, 13 F.4th at 973, as the video was virtually certain to contain nothing else of significance. Adopting the Maher/Wilson reasoning would create an unworkable, subjective, after-the-fact standard to afford a defendant a privacy interest in the details of CSAM even though they have no privacy interest in that file. This approach is at odds with Jacobsen and the private search doctrine.

¶43 As in Jacobsen, the test remains whether there is a virtual certainty that the government will not find anything of significance beyond what the private search revealed. Law enforcement may more thoroughly review the video, but if law enforcement confirms that the video Snapchat scanned, flagged, and reported is CSAM and nothing more, the fact that a person in law enforcement is the first to view the video does not equate to a private search being exceeded. And, Gasper does not argue that law enforcement viewed more videos or his entire account, or that law enforcement learned anything more from viewing the video. His sole argument is that the government exceeded Snapchat's search because a person in the government was the first to actually view the video.

¶44 Yet, Gasper urges that we require law enforcement to obtain a warrant before it can view what it is given by a private party. But, allowing law enforcement to view a tip, which here includes a video scanned and flagged as CSAM by a private party, before conducting a full-blown search, also makes practical and constitutional sense.²² The detective's review of the CSAM video allows law enforcement the opportunity to determine whether what the private party saw even warrants a more thorough investigation.

¶45 In short, law enforcement may receive a tip from any number of sources. The Fourth Amendment does not protect Gasper from a private actor who discovers that CSAM has been uploaded to its platform, discovered through the private actor's scan, and the private actor forwards that CSAM to the authorities. The government did not conduct a warrantless search of Gasper's cell phone or his Snapchat account; it merely reviewed the full CyberTip which included the video flagged as “apparent” CSAM. The private search doctrine applies when, such as here, a private actor invites a government agent to recreate the private actor's search and there is “virtual certainty that nothing else of significance” is in the file, and that inspection “would not tell [the State] anything more than [it] already had been told.” Jacobsen, 466 U.S. at 119, 104 S.Ct. 1652. Nothing about this review of a tip gave law enforcement unfettered access to Gasper's Snapchat account or his cell phone. Viewing the file was not a search that expanded that of Snapchat's.²³

¶46 “Under the private-search doctrine, the government does not conduct a Fourth Amendment search where there is a ‘virtual certainty’ that its search will disclose nothing more than what a private party's search has revealed.” Miller, 982 F.3d at 417–18 (quoting Jacobsen, 466 U.S. at 119, 104 S.Ct. 1652). The hash-based search of the files using Snapchat's digital eyes made it virtually certain that the detective would discover no more than what Snapchat had learned when a person in government viewed the images with human eyes. Jacobsen, 466 U.S. at 119, 104 S.Ct. 1652. Here, because the government “viewed only files with hash-value matches[,] ․ the private-search doctrine applies.” Miller, 982 F.3d at 429. In other words, Gasper's individual file was inspected and deemed suspicious by a private actor, and the private actor flagged the single, 16-second video and provided it to NCMEC. Accordingly, whatever expectation of privacy Gasper might have had in the hash values of his file was frustrated by Snapchat's private search. Opening the file “merely confirmed that the flagged file was indeed [CSAM], as suspected.” Reddick, 900 F.3d at 639.

C. Gasper's Reasonable Expectation of Privacy

¶47 Gasper argues that he has a categorical expectation of privacy in the reported video under Riley, 573 U.S. 373, 134 S.Ct. 2473, 189 L.Ed.2d 430. Under the facts of Riley, the Supreme Court held that warrantless searches of cell phones are presumptively unreasonable. 573 U.S. at 403, 134 S.Ct. 2473. Gasper argues that he had a “ ‘reasonable expectation of privacy’ in the cyberdata uploaded from his cellphone to his Snapchat account” and in the “content extracted” from that account. Gasper posits that because he accessed Snapchat exclusively from his cell phone, this case should be analyzed as if it was a cell phone search, and that he has a reasonable expectation of privacy of his cell phone and the Snapchat account.²⁴

¶48 Because we conclude that Snapchat performed a private search when it scanned and identified the flagged video as CSAM and the government did not exceed the scope of Snapchat's private search when it viewed the video, we need not analyze whether Gasper possessed a reasonable expectation of privacy entitling him to Fourth Amendment protection. We also need not determine to what extent, if any, Snapchat's terms of service agreement influences a reasonable expectation of privacy in a Fourth Amendment analysis.

D. Good-Faith Exception

¶49 Finally, Gasper argues that the good-faith exception to the exclusionary rule does not apply because that would swallow the warrant requirement. Because we determine that a warrant is not required in the case at issue, we need not analyze the good-faith exception. We caution, however, that “[t]o trigger the exclusionary rule, police conduct must be sufficiently deliberate that exclusion can meaningfully deter it, and sufficiently culpable that such deterrence is worth the price paid by the justice system.” Herring v. United States, 555 U.S. 135, 144, 129 S.Ct. 695, 172 L.Ed.2d 496 (2009). The conduct must be deliberate, reckless, or grossly negligent or the result of recurring or systemic negligence. Id.

¶50 We need not consider the parties’ arguments about the good-faith exception further, because we conclude that the search was not unconstitutional.

IV. CONCLUSION

¶51 The Fourth Amendment serves as a limit on government power. See Camara, 387 U.S. at 528, 87 S.Ct. 1727 (“The basic purpose of this Amendment, as recognized in countless decisions of this Court, is to safeguard the privacy and security of individuals against arbitrary invasions by government officials.”); Carpenter, 585 U.S. at 305, 138 S.Ct. 2206 (explaining that the Fourth Amendment places hurdles “in the way of a too permeating police surveillance”). A private search is not a government search. Ginglen, 467 F.3d at 1074. The Fourth Amendment is inapplicable to a search which has been completed by a private party as that search frustrates an individual's expectation of privacy. Jacobsen, 466 U.S. at 117, 104 S.Ct. 1652. The Fourth Amendment is implicated, however, if the government exceeds the private search. Id. at 115–22, 104 S.Ct. 1652. Gasper does not argue that the government viewed more than the one video provided, nor does he argue that anything else of significance was in the video. Gasper relies entirely on the argument that the government exceeded Snapchat's private search because a person in the government was the first to open and view the video, and did so without a warrant.

¶52 We conclude that the private search doctrine applies. It is undisputed that Snapchat performed a private search when it scanned and flagged the single, 16-second video as CSAM. The government did not exceed the scope of Snapchat's search when it viewed the video because any expectation of privacy Gasper may have had in the video was frustrated by the private search, and there was virtual certainty that law enforcement would not find anything of significance beyond what the private search revealed. As a result, the Fourth Amendment is not implicated. Accordingly, we affirm the court of appeals and remand to the circuit court for further proceedings consistent with this opinion.

By the Court.—The decision of the court of appeals is affirmed, and this cause is remanded to the circuit court for further proceedings consistent with this opinion.

¶53 I join the opinion that I wrote for the majority and write separately to expound upon this area of the law.

I. BACKGROUND: FEDERAL LAW REQUIREMENTS

¶54 Although we address the Fourth Amendment's application to Wisconsin's state law in the majority opinion, we did not detail the federal law implications when it comes to reporting CSAM. See United States v. Miller, 982 F.3d 412, 424 (6th Cir. 2020). Though Snapchat is not required to use any particular technology to identify CSAM, if it finds CSAM, then federal law requires it to report that CSAM to NCMEC. 18 U.S.C. § 2258A(a), (f). If such a CyberTip is forwarded to NCMEC, then NCMEC must forward the CyberTip to law enforcement for investigation. 18 U.S.C. § 2258A(c). The stated purpose of these laws is “to reduce ․ and ․ prevent the online sexual exploitation of children.” 18 U.S.C. §§ 2258A(a)(1)(A)(i), (a)(2)(A), 2510(15), 2258E.

¶55 Federal law requires Snapchat to report when it becomes aware of “apparent violations of [CSAM].” 18 U.S.C. § 2258A(a). That is exactly what Snapchat did. Snapchat reported the video and stated that it contained “apparent [CSAM].” Snapchat arrived at this conclusion utilizing its hash-value algorithm—such a system has been deemed reliable, akin to a digital fingerprint, a DNA match, or a VIN number. United States v. Ackerman, 831 F.3d 1292, 1294 (10th Cir. 2016); United States v. Miller, 982 F.3d 412, 418 (6th Cir. 2020); see also United States v. Dunning, 2015 WL 13736169, at *2 (E.D. Ky. Oct. 1, 2015) (finding that the chance of two of these files coincidentally sharing the hash value to be one in 9.2 quintillion—that is, highly unlikely). Snapchat's algorithm viewed each, individual pixel of the image, compared it to a database filled with known CSAM images, and determined that it contained contraband. Then, that single, 16-second video was removed from Gasper's account, and Snapchat followed the procedure outlined in federal law. In this case, Snapchat alone decided that this video contained CSAM. Snapchat, as a private actor, followed the letter of the federal law. Thus, federal law supports the process used in this case. Once Snapchat reported the CSAM, the government did not expand the scope of the search. It was Snapchat who frustrated Gasper's expectation of privacy—not the government.

¶56 Snapchat, through its algorithm, used its own resources to search for and identify contraband. Like the Supreme Court in United States v. Jacobsen, 466 U.S. 109, 104 S.Ct. 1652, 80 L.Ed.2d 85 (1984), other federal appellate courts have also applied the private search doctrine and concluded that there is no expansion of the private search when there is a “virtual” or “substantial” certainty that the government agent's search will not reveal anything more than what the private party represented. See United States v. Phillips, 32 F.4th 865, 870 (9th Cir. 2022); United States v. Rivera-Morales, 961 F.3d 1, 11, 15 (1st Cir. 2020); United States v. Lichtenberger, 786 F.3d 478, 488 (6th Cir. 2015); Rann v. Atchison, 689 F.3d 832, 836–37 (7th Cir. 2012). However, in Jacobsen, unlike the case at issue, law enforcement's search exceeded the search conducted by the Federal Express employees. Law enforcement opened the box, observed the baggies of powder, and tested the substance for cocaine. Although the testing was clearly beyond the employee's private search, the court concluded that the intrusion was nonetheless de minimis. As such, Jacobsen teaches that virtual certainty does not necessarily mean identical. Once the private search has frustrated an individual's reasonable expectation of privacy, the Fourth Amendment does not always require that the private search be perfectly replicated by the government.

¶57 Quite obviously, law enforcement is not required to avert its eyes from criminal activity. Viewing the provided video allowed law enforcement to confirm or dispel that it contained CSAM, as reported. Here, the government viewed what Snapchat's private search revealed: one, 16-second CSAM video from Gasper's account. It viewed what Snapchat provided and nothing more. Foregoing a warrant to view what this private actor provided through its own private search, is not only practical, it is constitutional.

¶58 Snapchat followed federal law by reporting the flagged video to NCMEC, and then NCMEC carried out its duties by forwarding the CyberTip to the government. In other words, this “apparent violation” followed congressional safeguards, which exist to reduce and prevent online sexual exploitation of children. 18 U.S.C. § 2258A(a).¹

II. PRACTICAL IMPLICATIONS: WARRANT

¶59 Gasper would require the detective to first obtain a warrant to view the CSAM-video Snapchat provided. But Gasper's argument assumes Fourth Amendment protection exists here, despite this being a private search. He also assumes that law enforcement would limit its warrant request to the video alone. As the majority opinion explained, the government is not required to obtain a warrant before viewing this private search. And as a practical matter, if law enforcement had probable cause to obtain a warrant to view the video for CSAM, then it likely follows that it would have probable cause to seek a much broader warrant searching Gasper's entire account, home, and electronic devices. Even the Second Circuit in United States v. Maher, a case that required a warrant before viewing CSAM, acknowledged that CyberTips can establish probable cause for a broader search warrant. See Maher, 120 F.4th 297, 319 (2d Cir. 2024) (stating that the CyberTip would have “demonstrate[d] probable cause to support warrants for ․ searches of ․ Google accounts and residence[s].”); see also United States v. Cartier, 543 F.3d 442, 446 (8th Cir. 2008). Given that the Fourth Amendment's touchstone is reasonableness, it is more reasonable for law enforcement to conduct this limited review of a private search before engaging in a far more invasive investigation based on an expansive warrant. State v. Tullberg, 2014 WI 134, ¶29, 359 Wis. 2d 421, 857 N.W.2d 120.

III. THE GOOD-FAITH EXCEPTION

¶60 To be clear, “[t]o trigger the exclusionary rule, police conduct must be sufficiently deliberate that exclusion can meaningfully deter it, and sufficiently culpable that such deterrent is worth the price paid by the justice system.” Herring v. United States, 555 U.S. 135, 144, 129 S.Ct. 695, 172 L.Ed.2d 496 (2009). The conduct must be “deliberate, reckless, or grossly negligent or the result of ‘recurring or systemic negligence.’ ” Id. When police act in good faith, or an area of the law is unsettled, there is no police misconduct to deter. State v. Scull, 2015 WI 22, ¶44, 361 Wis. 2d 288, 862 N.W.2d 562; United States v. Dorosheff, 110 F.4th 999, 1004-05 (7th Cir. 2024), reh'g denied, No. 22-2291, 2024 WL 4178484 (7th Cir., Sept. 12, 2024). That is exactly what is missing here: a deterrent effect.

IV. VICTIMS’ RIGHTS

¶61 Lastly, while not dispositive in this case, it is interesting to note that Gasper's arguments highlight the conflict between who might have competing privacy interests in the CSAM. Wisconsin has long held strong victim rights protection both in the form of legislation and in our constitution. Wis. Const. art. I, § 9m; Wis. Stat. ch. 950. One might opine about the child victim's privacy interest in a CSAM file. Justice Sotomayor has recognized, “[t]here is little doubt that the possession of images of a child being sexually abused would amount to an intentional invasion of privacy tort—and an extreme one at that.” Paroline v. United States, 572 U.S. 434, 483, 134 S.Ct. 1710, 188 L.Ed.2d 714 (2014) (Sotomayor, J., dissenting).

¶62 Consider also, Wisconsin's statutory right of privacy. Wisconsin Stat. § 995.50 states that a person's privacy is protected regardless of whether there is a criminal action. That section specifically references Wis. Stat. § 942.09 with respect to “intimate representations.” It could be that a minor would be deemed “incapable of consent” under § 942.09(1)(ae). We save these considerations for another day.

¶63 I would note, however, that Gasper's arguments that a person at Snapchat needed to view the CSAM before law enforcement could, would subject a child victim to even more victimization. The more human eyes that witness the wrongdoing, the more the child is victimized. And, if employees of the ESPs are forced, under Gasper's logic, to personally view and witness the CSAM video, they too may be victims of secondary trauma.² But for now, those considerations will be left for another day.

V. CONCLUSION

¶64 For all the foregoing reasons, I respectfully concur.

¶65 Now more than ever we live in a digital world. Vast amounts of information are created, stored, and shared using smartphones, computers, and other digital devices. And those devices “are portals to an endless array of online services [and] communities” where we can store our private information or share it with friends and strangers alike, all with the tap of a finger. Orin Kerr, The Digital Fourth Amendment 2 (2025). Now that we “live an online existence that can rival the physical one,” courts must grapple with the difficulties of adapting existing Fourth Amendment rules in order to preserve the delicate balance between privacy rights and the needs of law enforcement. Id. at 2–5.

¶66 Unfortunately both the court of appeals and majority fail at that task in this case, weakening our Fourth Amendment rights in the process. The court of appeals did so by concluding that boilerplate terms of service imposed by electronic service providers like Snapchat can limit or even eliminate users’ Fourth Amendment rights online. See, e.g., State v. Gasper, 2024 WI App 72, ¶¶21–22, 414 Wis. 2d 532, 16 N.W.3d 279. And although the majority wisely vacates the court of appeals’ published opinion, it does so only summarily. See majority op., ¶6, n.6. Making matters worse, the majority also misapplies the private-search doctrine, and concludes that no Fourth Amendment violation occurred in this case. I write separately to explain why both of these decisions are wrong, and why the good-faith exception to the exclusionary rule nonetheless applies.

I

¶67 The Fourth Amendment guarantees the right of the people to be free in “their persons, houses, papers, and effects [from] unreasonable searches and seizures ․” U.S. Const. amend. IV. By its terms, the Fourth Amendment therefore applies only to searches and seizures by the government, and only if they are unreasonable. A “search” in this context is a government intrusion into an area or object in which an individual has a reasonable expectation of privacy. See Kyllo v. United States, 533 U.S. 27, 32–33, 121 S.Ct. 2038, 150 L.Ed.2d 94 (2001).

¶68 In this case, the State claims that Gasper lacked a reasonable expectation of privacy in the video he privately uploaded to his account because he agreed to and subsequently breached Snapchat's terms of service.¹ If that were true it would be dispositive, since “no ‘search’ ․ occur[s] for Fourth Amendment purposes” if “the person objecting to a government intrusion lacks a reasonable expectation of privacy in the area examined ․” United States v. Shelton, 997 F.3d 749, 758 n.2 (7th Cir. 2021).

¶69 The court of appeals agreed with the State's argument, holding that the Fourth Amendment was not implicated when the DOJ analyst viewed the video for the first time without obtaining a warrant because Gasper lacked a reasonable expectation of privacy in the video. See Gasper, 414 Wis. 2d 532, ¶28, 16 N.W.3d 279. In doing so, the court of appeals determined that the private terms-of-service agreement between Gasper and Snapchat eliminated any Fourth Amendment protection Gasper had in the video vis-à-vis the government. See id. This holding, however, is a significant departure from settled Fourth Amendment doctrine in analogous real-world contexts, and would result in lesser Fourth Amendment protections in the digital world.

¶70 Before explaining why, it is helpful to describe what terms of service are. To create an account with an electronic service provider like Snapchat, users must agree to the terms of service, that is, “contractual language giving the company broad rights over” users’ accounts and the files stored there. See Orin Kerr, Terms of Service and Fourth Amendment Rights, 172 U. Pa. L. Rev. 287, 289 (2024). Two types of provisions are commonplace. The first are what Professor Orin Kerr calls “breach provisions,” which “explain what the company considers a breach that allows the company to limit or delete the user's account.” Id. at 292. And the second are “rules-of-the-road provisions,” which “set expectations about how a service will be run, such as what the company will do with [users’] data in various circumstances.” Id.

¶71 The court of appeals relied on both types of provisions when it concluded that Gasper lacked a reasonable expectation of privacy in the video.² Specifically, it relied on breach provisions, like Snapchat's prohibition on uploading “nude or sexually explicit content involving anyone under the age of 18.”³ Gasper, 414 Wis. 2d 532, ¶18, 16 N.W.3d 279. The court of appeals also cited rules-of-the-road statements by Snapchat, including that it could “access, review, screen, and delete” user-uploaded content at any time for any reason, and that it would “report all instances of child sexual exploitation to authorities ․” See id., ¶¶17–18. Together, the court of appeals held, these provisions in the terms of service meant Gasper lacked a reasonable expectation of privacy in the video (1) because uploading the video to his account breached the prohibition on uploading child-sexual-abuse material (CSAM) to the platform; and (2) because the rules-of-the-road provisions put him on notice that Snapchat could access, review, screen, or delete his content and would report any CSAM it found to law enforcement. Id. ¶¶21–22, 28.

¶72 The court of appeals’ decision and others like it ⁴ all rest on the false premise that in the digital world, the terms of private agreements and breaches of those terms can curtail or even eliminate expectations of privacy against the government. Courts have rejected that premise across a variety of analogous real-world contexts, however, and rightly held that private contracts “have little or no effect on Fourth Amendment rights.” Kerr, Terms of Service, supra at 308.

¶73 Car-rental contracts, apartment leases, and hotel-rental agreements are real-world counterparts to terms of service in the digital world. Like terms of service, each of these agreements allows a private party to use an owner's property subject to contractual limitations. Yet breaching provisions in a car-rental contract, even ones that specify that they void the agreement, does not result in an unauthorized driver losing their reasonable expectation of privacy in the vehicle. See United States v. Byrd, 584 U.S. 395, 408, 138 S.Ct. 1518, 200 L.Ed.2d 805 (2018). Likewise, breaches of apartment leases and hotel-rental agreements do not extinguish renters’ reasonable expectation of privacy in their apartment or hotel room. See, e.g., United States v. Thomas, 65 F.4th 922, 923–25 (7th Cir. 2023); United States v. Cunag, 386 F.3d 888, 895 (9th Cir. 2004). To be sure, breaches of these agreements may lead to eviction. But “the right to [evict] does not imply a right to [invite police to search the residence].” Thomas, 65 F.4th at 924; but see State v. Whitrock, 161 Wis. 2d 960, 966, 975–76, 468 N.W.2d 696 (1991) (concluding that a landlord could consent to a search after serving notice of eviction and believing the tenant had vacated the premises).

¶74 Granting a contractual right of access to an otherwise private space in the real world similarly does not eliminate reasonable expectations of privacy, thus authorizing law enforcement to access that space without a warrant. Apartment leases and hotel-rental agreements commonly include terms permitting the apartment owner or manager to access the unit for inspections or maintenance, or allowing hotel management or housekeeping to enter a guest's room for maintenance or cleaning. Yet in both contexts, courts have made clear that granting such a right of access doesn't eliminate the renter's reasonable expectation of privacy and open the space up to warrantless government searches. See United States v. Warshak, 631 F.3d 266, 287 (6th Cir. 2010).

¶75 Nor does it matter for purposes of the reasonable-expectation-of-privacy analysis that private parties might use their contractual right to access an otherwise private space to uncover information and share it with the government. “It is true, of course, that sharing space creates risks that a co-occupant will share [otherwise private] information with the government.” See Kerr, Terms of Service, supra at 307. But just because the government could discover information through someone else does not mean the government can enter a private space directly and take the information itself. See, e.g., State v. Bowers, 2023 WI App 4, ¶22, 405 Wis. 2d 716, 985 N.W.2d 123 (collecting cases). That is why the United States Supreme Court concluded, for example, that a warrantless search of a shared office at a union local violated the Fourth Amendment. See Mancusi v. DeForte, 392 U.S. 364, 369, 88 S.Ct. 2120, 20 L.Ed.2d 1154 (1968). Even though the defendant shared the office with others, and thus did not have a reasonable expectation of privacy with respect to those individuals, he still had a reasonable expectation of privacy against a warrantless search of that office by the government. See id.

¶76 These same principles should apply with equal force in the digital setting of this case. See State v. Baric, 2018 WI App 63, ¶19, 384 Wis. 2d 359, 919 N.W.2d 221 (emphasizing that “the reasonableness of an expectation of privacy in digital files ․ on electronic platforms is determined by considering the same factors as in any other Fourth Amendment context”). Because breaches of private agreements in the real world do not eliminate renters’ or users’ reasonable expectations of privacy against government intrusion, it is irrelevant that Gasper breached Snapchat's terms of service when he privately uploaded the video to his account. Compare Byrd, 584 U.S. at 408, 138 S.Ct. 1518; Thomas, 65 F.4th at 923–25; and Cunag, 386 F.3d at 895; with Gasper, 414 Wis. 2d 532, ¶¶18, 21–25, 16 N.W.3d 279. Snapchat's contractual right to, for example, delete his account or the video for breaching the terms of service says nothing about whether the government could search the video without a warrant. See Thomas, 65 F.4th at 924. And the fact that Snapchat's terms stated it could search Gasper's account and would report CSAM it found to law enforcement is similarly irrelevant. See Gasper, 414 Wis. 2d 532, ¶¶17–18, 16 N.W.3d 279. That is because granting a private party the contractual right to access an otherwise private space doesn't mean the government can access that same space without first obtaining a warrant. See Warshak, 631 F.3d at 287; Bowers, 405 Wis. 2d 716, ¶22, 985 N.W.2d 123. Thus, even though the terms of service put Gasper on notice that Snapchat might turn over his files to the government, “the mere ability of a third-party intermediary to access the contents of a [file] cannot be sufficient to extinguish a reasonable expectation of privacy.” See Warshak, 631 F.3d at 286 (emphasis in original).

¶77 The court of appeals’ holding to the contrary would severely undermine individuals’ privacy online. After all, if an electronic service provider's terms of service can eliminate a user's reasonable expectations of privacy in their digital files, then that means the government is free to access those files without obtaining a warrant and without implicating the Fourth Amendment. See United States v. Jones, 565 U.S. 400, 406, 132 S.Ct. 945, 181 L.Ed.2d 911 (2012).

¶78 If that sounds alarming, that's because it is. Social-media platforms like Snapchat are an omnipresent part of modern society. Over the last few decades, these platforms have transformed communication, supplanting older technologies. On Snapchat, users can send photos, videos, or messages instantly to friends, family, or strangers around the world. Those messages can contain anything from intimate private details about a user's life to funny cat videos. And that is equally true on Facebook, Instagram, and countless other social-media sites. If the voluminous, highly detailed, and broad terms of service imposed by these sites as a condition of creating an account “can narrow or eliminate Fourth Amendment rights online, then those rights may be an illusion. What the Supreme Court has given, [t]erms of [s]ervice might take away.” Kerr, Terms of Service, supra at 289.

¶79 Correctly understood, however, terms of service have little or no relevance to Fourth Amendment rights. They have at best limited bearing on consent to search. See State v. Kieffer, 217 Wis. 2d 531, 542, 577 N.W.2d 352 (1998) (describing the third-party consent doctrine). After all, the right to consent depends not “upon the law of property”—something the terms might demonstrate—but instead on actual shared use and control, which terms of service cannot establish on their own. See United States v. Matlock, 415 U.S. 164, 171 n.7, 94 S.Ct. 988, 39 L.Ed.2d 242 (1974). Likewise, terms of service are only tangentially relevant to the private-search doctrine. For example, terms of service might be a minor point on the scale, helping to show whether an electronic service provider was acting as a private party or an agent of the government when it searched a user's account. See United States v. Rosenow, 50 F.4th 715, 730 (9th Cir. 2022). But as with consent to search, the application of the private-search doctrine hinges on far more than just terms of service. See generally id. at 728–35.

¶80 What terms of service cannot do, however, is eliminate or even limit a user's reasonable expectations of privacy online vis-à-vis a government search. To hold otherwise, as the court of appeals did in this case, is to make citizens’ Fourth Amendment rights online rise and fall on the whim of tech companies and large corporations. These important rights are, and must remain, more resilient than that.

II

¶81 Although the majority rightly vacates the court of appeals’ published opinion, unfortunately its decision erodes Fourth Amendment rights in different way. As Justice Crawford's separate writing correctly explains, Gasper's Fourth Amendment rights were violated when, without obtaining a warrant, a Wisconsin DOJ analyst viewed for the first time a video privately uploaded to Gasper's Snapchat account. In concluding otherwise, the majority misapplies the private-search doctrine, holding that it was “virtual[ly] certain[ ]” that by viewing the video for the first time, the analyst “would not find anything of significance beyond what” was already revealed by Snapchat's prior private search. See majority op., ¶6; see also United States v. Jacobsen, 466 U.S. 109, 119, 104 S.Ct. 1652, 80 L.Ed.2d 85 (1984). But that private search was limited in scope, scanning only the video's hash value—“a sort of digital fingerprint”⁵ for computer files—and identifying it as a match for the hash value of an image or video previously flagged as containing CSAM. The hash match alone, however, “revealed nothing, either to [Snapchat] or those with whom it shared the match, about what in particular the [video] depicted (or even what the [file it matched to] depicted).” United States v. Maher, 120 F.4th 297, 306 (2d Cir. 2024). And for that reason, the analyst's actions exceeded the scope of Snapchat's prior private search.

¶82 In arguing otherwise, Justice Hagedorn's concurrence illustrates the pitfalls inherent in analogizing Snapchat's hash-value search to other contexts. To begin with, he relies on a real-world case in which individuals took private documents and turned them over to law enforcement. See Justice Hagedorn's concurrence, ¶100 (citing Burdeau v. McDowell, 256 U.S. 465, 475, 41 S.Ct. 574, 65 L.Ed. 1048 (1921)). But in that case, law enforcement didn't have to open a sealed envelope or other container to view the documents, their contents were plain for anyone to see.⁶ See United States v. Knoll, 16 F.3d 1313, 1320–21 (2d Cir. 1994) (explaining that a warrant is required for law enforcement to examine stolen files sealed in folders and boxes). That is not true in this case, however, where the only way to know the complete contents of the digital video was to open the file and view it.

¶83 Nevertheless, Justice Hagedorn's concurrence relies on this real-world case in comparing Snapchat's actions to a private party's hypothetical “keyword search of ․ emails” for the name of a bank with known ties to organized crime. See Justice Hagedorn's concurrence, ¶99. The hypothetical keyword search turns up five emails, which the searcher reads enough to confirm “they all contain the name of the shady bank.” Id. The searcher then prints them out, and hands them over to the police. See id. What this example proves is anyone's guess, however, since the hypothetical bears virtually no resemblance to the facts of this case. Unlike the concurrence's hypothetical searcher, who visually examined each email to confirm it contained the potentially incriminating information (thus invading any expectation of privacy the account holder might have had), no one at Snapchat viewed the video before it was opened by law enforcement. Moreover, unlike the unopened video file attached to the CyberTip, the hypothetical printed-out emails revealed their contents for anyone to see. In the end, all this hypothetical demonstrates is the problem with relying too heavily on real-world analogies in the digital context. As the United States Supreme Court has cautioned, such an approach results at best in “a difficult line-drawing expedition to determine which digital files are comparable to physical records,” and at worst in “a significant diminution of privacy.” See Riley v. California, 573 U.S. 373, 400–01, 134 S.Ct. 2473, 189 L.Ed.2d 430 (2014).

¶84 Perhaps for that reason, Justice Hagedorn's concurrence abandons this argument to articulate a “second way this case can be resolved,” namely “by following the analysis in the field test portion of [United States v.] Jacobsen.” Justice Hagedorn's concurrence, ¶104. In Jacobsen, the United States Supreme Court held that even though DEA agents exceeded the scope of a prior private search when they performed a field test on white powder discovered in a package, that test did not violate the Fourth Amendment. See 466 U.S. at 123, 104 S.Ct. 1652. That was so, the Court explained, because the binary nature of the test, which could reveal only whether the powder was or was not cocaine, “does not compromise any legitimate interest in privacy.” Id. The same result should follow here, Justice Hagedorn's concurrence asserts, because “[a]ny additional invasion of Gasper's reasonable expectation of privacy in this video was small to non-existent” since “its contents had already been searched and it had already been flagged for illegal CSAM.” Justice Hagedorn's concurrence, ¶104.

¶85 This reading of Jacobsen is a novel one, in that it treats Snapchat's actions—limited in scope as they were—as conclusive of whether Gasper had any remaining expectation of privacy in the video. But Jacobsen did not focus on what the FedEx employees did, or the scope of their search. Instead, the Court focused on the fact that the field test could reveal only “whether a substance is cocaine, and no other arguably ‘private’ fact.” See Jacobsen, 466 U.S. at 123, 104 S.Ct. 1652. Accordingly, the federal courts have, with one exception, rejected arguments like the one Justice Hagedorn's concurrence makes.⁷ Visually examining a file flagged as containing CSAM reveals all of the contents of that video, and thus “is a far cry from a field test's disclosure of nothing more than a binary answer.” Maher, 120 F.4th at 316. Such a visual examination “reveals innumerable granular private details,” and thus “implicat[es] privacy interests beyond a binary classification.” United States v. Wilson, 13 F.4th 961, 979 (9th Cir. 2021). Therefore, I conclude that Gasper still had a reasonable expectation of privacy in the video even after Snapchat's private search, and join the portion of Justice Crawford's separate writing explaining why the events in this case violated the Fourth Amendment. See Justice Crawford's concurrence in part and dissent in part, ¶¶113–24.

III

¶86 When evidence is obtained in violation of the Fourth Amendment, the typical remedy is to exclude evidence obtained through that unlawful search. See State v. Blackman, 2017 WI 77, ¶68, 377 Wis. 2d 339, 898 N.W.2d 774. Nevertheless, the exclusionary rule is not automatic, and should be applied only when doing so would yield “appreciable deterrence.”⁸ Davis v. United States, 564 U.S. 229, 237, 131 S.Ct. 2419, 180 L.Ed.2d 285 (2011) (quoting another source).

¶87 To that end, the Supreme Court has applied the “good-faith exception” to the exclusionary rule, which recognizes that when law enforcement's conduct is less culpable, applying the exclusionary rule is less likely to lead to meaningful deterrence. Id. at 238 (quoting Herring v. United States, 555 U.S. 135, 143, 129 S.Ct. 695, 172 L.Ed.2d 496 (2009) (internal alteration omitted)). If law enforcement acts “in objectively reasonable reliance on ․ a facially valid warrant properly issued by a neutral, detached magistrate; an apparently constitutional statute; or a binding appellate precedent,” applying the exclusionary rule is not warranted. State v. Burch, 2021 WI 68, ¶79, 398 Wis. 2d 1, 961 N.W.2d 314 (Dallet, J., concurring in part, dissenting in part) (citing United States v. Leon, 468 U.S. 897, 104 S.Ct. 3405, 82 L.Ed.2d 677 (1984); Illinois v. Krull, 480 U.S. 340, 107 S.Ct. 1160, 94 L.Ed.2d 364 (1987); Davis, 564 U.S. at 239–41, 131 S.Ct. 2419). By contrast, when law enforcement demonstrates “ ‘deliberate,’ ‘reckless,’ or ‘grossly negligent’ disregard for Fourth Amendment rights,” the exclusionary rule should apply to deter that misconduct. See Davis, 564 U.S. at 238, 131 S.Ct. 2419 (quoting Herring, 555 U.S. at 144, 129 S.Ct. 695); see also Burch, 398 Wis. 2d 1, ¶80, 961 N.W.2d 314 (Dallet, J., concurring in part, dissenting in part).

¶88 Some situations, like the one in this case, fall between those two poles. And when that happens, courts must assess the situation's unique facts, weighing the costs of suppression against the deterrence benefits of exclusion in light of “the ‘flagrancy of the police misconduct’ at issue.” Davis, 564 U.S. at 238, 131 S.Ct. 2419 (quoting Leon, 468 U.S. at 909, 104 S.Ct. 3405). In other words, applying the good-faith exception requires much more than the broad, legally incorrect, and conclusory statement in Justice Ziegler's concurrence that “[w]hen police act in good faith, or an area of the law is unsettled, there is no police misconduct to deter.” Justice Ziegler's concurrence, ¶60.

¶89 Here's what the good-faith-exception analysis should look like. At the time the DOJ analyst viewed for the first time the video privately uploaded to Gasper's Snapchat account, the Fifth and Sixth Circuits held that a warrant was not required before doing so under the private-search doctrine. See, e.g., United States v. Miller, 982 F.3d 412, 426–34 (6th Cir. 2020); United States v. Reddick, 900 F.3d 636, 638–40 (5th Cir. 2018). Two state courts reached the same conclusion. See, e.g., People v. Wilson, 56 Cal.App.5th 128, 270 Cal. Rptr. 3d 200, 220–25 (2020); Morales v. State, 274 So. 3d 1213, 1217–18 (Fla. Ct. App. 2019). The Ninth Circuit disagreed, however, holding that a warrant was required.⁹ See Wilson, 13 F.4th at 964. According to testimony at the suppression hearing in this case, attorneys at DOJ analyzed this split of authority and concluded that a warrant was not required before law enforcement in Wisconsin opened for the first time a file allegedly containing CSAM that was attached to a CyberTip.

¶90 As I have written before, in the face of uncertainty, law enforcement should of course “ ‘err on the side of constitutional behavior’ and get a warrant.” Burch, 398 Wis. 2d 1, ¶83, 961 N.W.2d 314 (Dallet, J., concurring in part, dissenting in part) (quoting another source). Had they done so here, years of appellate proceedings could have been avoided at virtually no cost, since such a warrant would have been easy to obtain. Nevertheless, under these circumstances, I would not apply the exclusionary rule. Law enforcement's actions here were not the kind of “ ‘deliberate,’ ‘reckless,’ or ‘grossly negligent’ disregard for Fourth Amendment rights” the exclusionary rule is needed to deter. See Davis, 564 U.S. at 238, 131 S.Ct. 2419. Instead, law enforcement made a reasoned, though mistaken in my view, decision to follow the weight of non-binding authority on an unsettled legal question, as reflected in the decisions of two federal circuits and two state appellate courts. See id.; see also United States v. Ford, No 1:11-CR-42, 2012 WL 5366049, at *11 (E.D. Tenn. Oct. 30, 2012) (reaching a similar conclusion with respect to a 3-1 circuit split). Even though not all instances of law enforcement reliance on non-binding precedent will fall within the good-faith exception, this one does.

¶91 Accordingly, while I disagree with the majority's reasoning, I concur with its conclusion that Gasper's motion to suppress should have been denied. I therefore respectfully concur.

¶92 Snapchat digitally scans its users’ uploaded video content to see whether it contains known child sexual abuse material (CSAM). If it does, Snapchat flags the content and turns it over to law enforcement. In this case, Snapchat flagged a 16-second video as likely CSAM and forwarded it to law enforcement. When law enforcement received the video, they watched it—something no employee of Snapchat did. The main question in this case is whether law enforcement carried out an unreasonable search in violation of the Fourth Amendment when it watched the video. I agree with the majority that it did not. I write separately to add some additional context and analysis to this discussion.

¶93 Some basic Fourth Amendment principles guide our analysis. The Fourth Amendment generally prohibits a government search without a warrant when that search invades a reasonable expectation of privacy. United States v. Jacobsen, 466 U.S. 109, 113, 104 S.Ct. 1652, 80 L.Ed.2d 85 (1984). Searches performed by private actors, however, are not government searches and therefore not subject to Fourth Amendment protections. Id.

¶94 What happens, though, if a private actor conducts a search and then turns over evidence to law enforcement? Must the government get a warrant to examine what a private party has already searched and provided? The United States Supreme Court has said no; police need not “avert their eyes.” Coolidge v. New Hampshire, 403 U.S. 443, 489, 91 S.Ct. 2022, 29 L.Ed.2d 564 (1971). In Jacobsen, the Supreme Court set out a broader principle governing police searches of evidence that has been turned over by a private party: “The additional invasions of respondents’ privacy by the Government agent must be tested by the degree to which they exceeded the scope of the private search.” Jacobsen, 466 U.S. at 115, 104 S.Ct. 1652.

¶95 When a private actor conducts a search, the reasonable expectation of privacy has been frustrated. Id. at 117–18, 104 S.Ct. 1652. Thus, as long as the subsequent government search does not exceed the scope of the private search, the government has not invaded any additional expectation of privacy. When a search merely replicates what the private actor did, no Fourth Amendment search occurs. Id. at 119–20, 104 S.Ct. 1652. But Jacobsen also permits some searches that exceed the scope of the private party search if the additional infringement on the remaining interest in privacy is minimal. Id. at 124, 104 S.Ct. 1652. Where the residual privacy interest is negligible, Fourth Amendment interests similarly do not justify the need for a warrant. Id. at 123, 104 S.Ct. 1652.

¶96 Jacobsen applied this to two different searches by DEA (Drug Enforcement Administration) agents. For the first, DEA agents replicated what the private Federal Express employees had already done. They removed material found in the package that the employees had already opened and found a white powder. Id. at 111, 104 S.Ct. 1652. In replicating this search, the DEA agents weren't going to learn anything they didn't already know, so they did not infringe upon any further privacy interests that had not already been frustrated by the private search conducted by Federal Express employees. In the second search, DEA agents performed a field test on the white powder. Jacobsen initially observed that this search did exceed the scope of the prior search. Id. at 122, 104 S.Ct. 1652. But the search was still reasonable under the Fourth Amendment because the likelihood that any legitimate interest in privacy would be compromised was “much too remote” to implicate the Fourth Amendment. Id. at 124, 104 S.Ct. 1652. The additional invasion of privacy was small and unlikely to reveal any other “ ‘private’ fact.” Id. at 123, 104 S.Ct. 1652. Given the virtual certainty that the test would reveal only whether the white powder was cocaine, an illegal contraband, the search was reasonable under the Fourth Amendment even though it exceeded the scope of the private search. Following this reasoning, our case can be resolved in two ways consistent with Jacobsen.

¶97 First, as the majority explains, the government search did not exceed the scope of the private party's search. Snapchat, a private actor, conducted a digital search of the video—not just its label, but its contents. This means Gasper's expectation of privacy in the contents of the video were frustrated. A human search of the video may be different in form, but not in kind. It is a search of what was in the video—just like the one conducted by Snapchat. Therefore, when the government watched the video, it did not exceed the scope of Snapchat's private digital search. Moreover, it's not even clear that law enforcement's human viewing of the video should be thought of as any more invasive than the sophisticated search conducted by Snapchat—one that analyzes the video by comparing pixels within the video to a database of known CSAM.

¶98 Justice Crawford disagrees and argues that when a human viewed the video, the government exceeded the scope of Snapchat's search, analogizing this to a dog sniff alerting to narcotics in luggage at an airport. Justice Crawford's concurrence/dissent, ¶123. Just as a warrant would be required to search the luggage after the dog alerts, Justice Crawford reasons, a warrant is required after a digital CSAM alert before police may watch the video. Id. I respectfully disagree.

¶99 Snapchat's search was not external to the video in the way a dog sniff is external to luggage; it was a search of the video itself. The suitcase analogy falls short. A better way to view this is like a keyword search of emails. Suppose a woman suspects her husband is engaged in financial shenanigans and conducts a keyword search of his emails. She searches specifically for emails containing the name “Gambino Bank”—a local depository known for its ties to organized crime. The search locates five emails. She does not read the emails beyond her confirmation that they all contain the name of the shady bank. She then prints the emails and drops them off at the local police precinct, telling them that she believes these emails contain evidence that her husband is engaged in fraudulent financial activities.

¶100 The woman's private search in this hypothetical is digital and depends on the content of the emails. Can law enforcement, upon receipt of the documents, read them in full, or must they avert their eyes or obtain a warrant? Almost assuredly, courts would say law enforcement could read the emails. Indeed, in the seminal case establishing that the Fourth Amendment is not offended by a private party's search, the Supreme Court held that police can review documents turned over by private individuals and use them in a subsequent prosecution. Burdeau v. McDowell, 256 U.S. 465, 475–76, 41 S.Ct. 574, 65 L.Ed. 1048 (1921). The emails in my hypothetical have already been searched by the private actor, and reading the emails is no more invasive to a privacy interest than the keyword search, which likewise reviews each word and phrase. While law enforcement's examination of the incriminating emails may be different in form from the woman's keyword search, it is not different in kind.

¶101 The same logic should apply to the short video at issue here. Snapchat's digital search of the contents of the video is a real search, and law enforcement may permissibly search the contents of the video using a different method—here, watching it rather than conducting another digital scan. Under Jacobsen this is not an expanded search. Rather, given the frustration of any remaining privacy interests in the contents of the video, it remains within the scope of the private search.

¶102 One of the complications in this type of case is Jacobsen’s focus on what one might learn from a search, which is rooted in the Supreme Court's jurisprudence regarding the search of a container. Jacobsen focuses on whether the law enforcement's subsequent search is “virtually certain” to result in learning more than law enforcement already knew. It is not clear to me that this focus is equally illuminating in digital searches or when the potential criminal activity is more complex. I can imagine all kinds of circumstances where law enforcement is sure to learn more than a private actor even while conducting the exact same search. A private actor suspecting financial fraud, for example, is unlikely to notice all that a trained law enforcement officer would see when replicating the private party's search. And doctrinally, the private search doctrine rests upon whether the scope of the search has been exceeded, not whether law enforcement notices more than the private actor. See United States v. Simpson, 904 F.2d 607, 610 (11th Cir. 1990) (reasoning the government's search does not expand upon the private party's search “simply because they took more time and were more thorough than [the private party]”). For this reason, I'm not sure the emphasis on “virtual certainty” is as helpful in light of the kind of search we are examining here.

¶103 The real question from Jacobsen would seem to be whether the defendant's privacy interest was frustrated. Here, the whole video was searched by Snapchat, even though Snapchat did not have a human watch the video. In my view, the expectation of privacy in the video was frustrated by Snapchat's digital viewing of the video, which means law enforcement doesn't exceed the scope of the private search by also viewing the video—albeit in a different manner.

¶104 The second way this case can be resolved is by following the analysis in the field test portion of Jacobsen. To the extent this search exceeds the scope of the PhotoDNA hash search performed by Snapchat, we still must ask how much of a remaining expectation of privacy Gasper had in the video after Snapchat's search. The answer is not much. Gasper had little expectation of privacy remaining in the contents of this 16-second video after its contents had already been searched and it had already been flagged for illegal CSAM. Here, to the extent watching the video is deemed an additional government search by exceeding the scope of Snapchat's digital search, it isn't much of one. Any additional invasion of Gasper's reasonable expectation of privacy in this video was small to non-existent, and I would conclude it was insignificant—just like the field test in Jacobsen.

¶105 For these reasons, I respectfully concur.

¶106 Although the government's interests in protecting children from sexual abuse and exploitation and holding perpetrators accountable are unquestionably compelling, those interests do not excuse the government from following the basic commands of the Constitution. Here, the Fourth Amendment required the government to obtain a search warrant before opening and viewing Michael Gasper's Snapchat file.

¶107 Many electronic service providers (ESPs) digitally monitor their platforms for harmful content and voluntarily share suspected child sexual abuse materials (CSAM) with the government, as Snapchat did here. When the State opened and viewed the video file it received from Snapchat, it acquired information beyond what was detected by Snapchat's digital scan. The State utilized that additional information—a detailed description of the contents of the video—when it applied for a search warrant for Gasper's home and cell phone. The State should have, and readily could have, obtained a search warrant before viewing the video file it received from Snapchat. It chose not to do so. The State's deliberate decision to open and view the file without first obtaining a search warrant cannot be excused as good faith. I would affirm the circuit court's order suppressing the evidence the State obtained by opening and viewing the file, specifically the content of that video. I conclude, however, that the remaining facts gained from the CyberTip and investigation were sufficient to support probable cause for the search warrant of Gasper's home and devices. I thus agree with the mandate reversing the circuit court's order suppressing evidence obtained pursuant to the search warrant.

¶108 For these reasons, I concur in part and dissent in part.

I. BACKGROUND

¶109 This case represents an increasingly common fact pattern as courts grapple with the Fourth Amendment rights of citizens who store photos, videos, and other data in password-protected ESP accounts in “the cloud” (remote servers in data centers maintained by ESPs). In a routine scan, Snapchat's software detected potential CSAM in a video file Gasper had uploaded and saved to his personal Snapchat account. Without opening or viewing the flagged file, Snapchat emailed a CyberTip with the attached video to the National Center for Missing and Exploited Children (NCMEC). The CyberTip included the Snapchat user name, as well as the email address and IP address attached to the account. NCMEC determined that the device associated with the IP address was located in Wisconsin and was served by CenturyLink. NCMEC then emailed the CyberTip and the additional information to the Wisconsin Department of Justice (DOJ).

¶110 A DOJ employee opened and viewed the video attached to the CyberTip without obtaining a search warrant. After viewing the video, the DOJ employee obtained the name and address associated with the IP address from CenturyLink under an administrative subpoena. The employee forwarded the video file, along with Gasper's name and home address, to the Waukesha County Sheriff's Department (Sheriff's Department). Upon receipt, a detective viewed the video, again without first obtaining a search warrant. The detective then applied for and received a warrant to search Gasper's home and devices. The warrant affidavit included a detailed description of the content of the video to support a finding of probable cause.

II. REASONABLE EXPECTATION OF PRIVACY

¶111 A defendant challenging a search on Fourth Amendment grounds bears the burden of proving “that he or she had an actual, subjective expectation of privacy in the area search and item seized” and “that society is willing to recognize the defendant's expectation of privacy as reasonable.” State v. Tentoni, 2015 WI App 77, ¶7, 365 Wis. 2d 211, 871 N.W.2d 285.

¶112 The majority does not reach the question of whether Gasper had a reasonable expectation of privacy in the video file because it concludes that the State's opening of the file and viewing the video did not exceed the scope of Snapchat's “private search,” and thus does not implicate the Fourth Amendment. I disagree with the majority's conclusion that the State's actions did not exceed the scope of Snapchat's digital scan, as discussed below. I would also hold that the court of appeals erred in concluding that Gasper did not have a reasonable expectation of privacy in the video file because his “conduct was obviously unlawful” and contrary to Snapchat's terms of service. State v. Gasper, 2024 WI App 72, ¶22, 414 Wis. 2d 532, 16 N.W.3d 279. I agree with Justice Dallet that Gasper had a reasonable expectation of privacy in files he placed in his password-protected Snapchat account, and that Snapchat's specific terms of service did not extinguish that expectation of privacy. I join her concurrence on that issue. See Justice Dallet's concurrence, ¶¶67–80.

III. THE PRIVATE SEARCH DOCTRINE

¶113 “[T]he Fourth Amendment applies only to government action.” State v. Payano-Roman, 2006 WI 47, ¶17, 290 Wis. 2d 380, 714 N.W.2d 548. If the government repeats a search conducted by a private party, “[t]he Fourth Amendment is implicated only if the authorities use information with respect to which the expectation of privacy has not already been frustrated.” United States v. Jacobsen, 466 U.S. 109, 117, 104 S.Ct. 1652, 80 L.Ed.2d 85 (1984). Whether government agents have committed additional invasions of the defendant's privacy “must be tested by the degree to which they exceeded the scope of the private search.” Id. at 115, 104 S.Ct. 1652. The majority holds that the State did not exceed the scope of Snapchat's private search here. In their view, the DOJ and Sheriff's Department merely duplicated Snapchat's digital scan when they opened and viewed the video. But no Snapchat employee had viewed the video; nor did the government simply replicate Snapchat's digital scan of Gasper's file. If either had done so, this would be an easy case with no violation of the Fourth Amendment. Unfortunately, that's not what happened here.

¶114 Jacobsen held that when the government's inspection reveals “nothing else of significance” beyond what was disclosed to it by a private party, no legitimate privacy interest protected under the Fourth Amendment is infringed. Id. at 119, 104 S.Ct. 1652 (holding that a DEA agent's inspection of a package that FedEx employees had previously opened, disclosing bags of white powder, did not further infringe the defendant's privacy interests). The government here knew only that Snapchat had flagged the file as “apparent CSAM” after conducting a digital scan. It did not know what specimen of “known CSAM” Snapchat's scan had determined to digitally match Gasper's file. Nor did the government have a description of the content of the video. Until government agents opened and viewed the file, the State did not know if the file contained an intact video or if it depicted CSAM as defined by state law. See Richard P. Salgado, Fourth Amendment Search and the Power of the Hash, 119 Harv. L. Rev. F. 38, 40 (2005) (explaining that the hash value of a digital file “cannot be ‘reversed’ ” to reveal the content of the file itself). Only by opening and playing the file did the government confirm it contained an intact video that was unequivocally CSAM. The detective's inclusion of a detailed description of the video in the search warrant affidavit underscores the investigative value of the evidence obtained by opening the file. It is simply not true that “nothing else of significance” was disclosed by viewing the video.¹

¶115 Jacobsen teaches that the government's search does not exceed the scope of a private search if it did not learn anything beyond what it could have obtained from the private searcher's testimony. 466 U.S. at 118–20, 104 S.Ct. 1652; see also United States v. Runyan, 275 F.3d 449, 461 (5th Cir. 2001). That obviously is not the case here. The detective's search warrant affidavit, which includes a detailed description of the video, shows that the government obtained information useful to the prosecution by viewing the video. No Snapchat employee could have provided that information through testimony.

¶116 The majority incorrectly distinguishes the present case from Walter, an earlier case discussed at length in Jacobsen. See Walter v. United States, 447 U.S. 649, 651–52, 100 S.Ct. 2395, 65 L.Ed.2d 410 (1980). Employees opened packages misdelivered to a company, discovering boxes of films labeled with suggestive drawings and explicit descriptions of the contents. Id. An employee attempted to view the films by holding them up to the light, but was not successful. Id. Government agents viewed the films without a warrant. Id. The Court explained that “[p]rior to the Government screening one could only draw inferences about what was on the films.” Id. at 657, 100 S.Ct. 2395. As such, “[t]he projection of the films was a significant expansion of the search that had been conducted previously by a private party and therefore must be characterized as a separate search. That separate search was not supported ․ by a warrant even though one could have easily been obtained.” Id. Like the company employees in Walter, Snapchat's employees did not open or view the video file that its software flagged as “apparent CSAM.” Nor did NCMEC view the video. The video was not observable until government actors opened and played the file using an appropriate software program.² Cf. Walter, 447 U.S. at 652 n.2, 100 S.Ct. 2395 (explaining that the films could not “be examined successfully with the naked eye” due to their size). The majority attempts to distinguish Walter by stressing that it took federal agents months to review the hundreds of boxes of film. But neither the duration of the search nor the number of files examined have constitutional relevance. As in Walter, the government's viewing of the previously-unseen video expanded the private search, disclosing more information about the content of the file. Like the federal agents in Walter, the State could easily have obtained a search warrant before opening the file and viewing the video. It chose not to.

¶117 The federal courts differ on what the Fourth Amendment requires in cases like Gasper's, where the government opens and views a file flagged by an ESP's digital scan without first obtaining a search warrant. The Ninth and, most recently, the Second Circuit have held that the government's conduct is an unconstitutional expansion of the private search. See United States v. Wilson, 13 F.4th 961, 972 (9th Cir. 2021) (“[T]he government search ․ expanded the scope of the antecedent private search because the government agent viewed Wilson's email attachments even though no Google employee—or other person—had done so, thereby exceeding any earlier privacy intrusion.”); United States v. Maher, 120 F.4th 297, 320 (2d Cir. 2024) (“Because no one at Google had ever opened or visually examined the contents of the Maher file ․, such a visual examination by the police did not fall within the private search doctrine's exception to the warrant requirement.”).

¶118 The Fifth and Sixth Circuits have held that the private search doctrine does apply when government agents conduct a warrantless viewing of the contents of an unopened file attached to a CyberTip. See United States v. Reddick, 900 F.3d 636 (5th Cir. 2018); United States v. Miller, 982 F.3d 412 (6th Cir. 2020). They focus on the perceived reliability of the digital scan and gloss over the additional information government agents stand to gain by viewing the contents of the files. See Reddick, 900 F.3d at 639 (stating that viewing the file “merely dispelled any residual doubt about the contents of the files,” similar to the agent's field testing of the white powder in Jacobsen); Miller, 982 F.3d at 429–30 (stating that it was “virtually certain” viewing the files would reveal CSAM).

¶119 Unlike the majority, I find the federal cases holding that the government's conduct exceeds the private search to be persuasive and in alignment with Jacobsen. The digital scans conducted by ESPs provide only a binary determination that a file falls within a category of contraband, and even then, the classification is a tentative one: that a file is potential CSAM. Only the officers’ subsequent viewing of the video confirmed that the file contained intact CSAM and, in this case, disclosed the unambiguously illegal nature of the content under Wisconsin law. Cf. Wilson, 13 F.4th at 973 (“Until he viewed the images, they were at most ‘suspected’ child pornography․ Only by viewing the images did the government confirm, and convey to the fact finder in Wilson's criminal case, that they depicted child pornography under the applicable federal standard.”); Maher, 120 F.4th at 316 (“[A] human visual examination of a computer hash matched image does not disclose only whether or not the image depicts child pornography. Visual examination necessarily also reveals the particulars supporting either a ‘yes’ or ‘no’ answer.”).

¶120 Moreover, even assuming an ESP's hash-value scanning is highly reliable and accurate in identifying CSAM, that reliability does not dispense with the Fourth Amendment's warrant requirement. “[T]he reliability of [an ESP's] hash matching technology is pertinent to whether probable cause could be shown to obtain a warrant, not to whether the private search doctrine precludes the need for the warrant.” Maher, 120 F.4th at 319–20 (citation modified); see also United States v. Braun, 798 F. Supp. 3d 916, 925 (E.D. Wis. 2025) (quoting and relying on Maher in holding that a search warrant is required for officers to view an unopened file in a CyberTip). A law enforcement officer may, for example, have highly reliable information about the presence of drugs in a home; but the reliability of that information does not justify a warrantless entry and search. “Incontrovertible testimony of the senses that an incriminating object is on premises belonging to a criminal suspect may establish the fullest possible measure of probable cause. But even where the object is contraband, this Court has repeatedly stated and enforced the basic rule that the police may not enter and make a warrantless seizure.” Horton v. California, 496 U.S. 128, 137 n.7, 110 S.Ct. 2301, 110 L.Ed.2d 112 (1990).

¶121 The majority here, along with Justice Hagedorn's concurrence, similarly misapplies the Jacobsen Court's analysis of the federal agent's field testing of the white powder discovered by the FedEx employees. The Court conceded that the field test “exceeded the scope of the private search,” but held that the warrantless test did not compromise any legitimate expectation of privacy protected under the Fourth Amendment. 466 U.S. at 122–23, 104 S.Ct. 1652. The Court emphasized that the test would “merely disclose[ ] whether or not a particular substance [was] cocaine.” Id. at 123, 104 S.Ct. 1652. It implied that the contraband nature of the substance was already in plain view, stating that “[i]t is probably safe to assume that virtually all of the tests conducted under circumstances comparable to those disclosed by this record would result in a positive finding” and that it was “virtually certain” that the substance was contraband. Id. at 123, 125, 104 S.Ct. 1652. The Court thus focused its analysis not on the search, but the seizure: the test destroyed a trace amount of the powder, “convert[ing] what had been only a temporary deprivation of possessory interests into a permanent one.” Id. at 124–25, 104 S.Ct. 1652. It concluded that because the test had “a de minimis impact on any protected property interest,” and because the law enforcement interests were substantial, the field test was a reasonable seizure under the Fourth Amendment. Id. at 125, 104 S.Ct. 1652.

¶122 By contrast, the video contained in Gasper's file was not in view when the government received it; nor was viewing the video equivalent to the chemical testing of an obviously contraband white powder.³ “A visual examination's revelation of particulars is a far cry from a field test's disclosure of nothing more than a binary answer.” Maher, 120 F.4th at 316. See also Wilson, 13 F.4th at 978–79; Miller, 982 F.3d at 429 (concluding that the private search doctrine supported the government's warrantless viewing but rejecting the Fifth Circuit's reasoning from Reddick “that the detective's viewing of the images was like the DEA agent's testing of the powder in Jacobsen”). The Court's focus on the loss of property caused by testing the powder likewise has no parallel here.

¶123 The Jacobsen Court compared the field test to a trained canine alerting to the scent of narcotics in luggage at an airport, observing that both disclose “only the presence or absence of narcotics, a contraband item.” 466 U.S. at 123–24, 104 S.Ct. 1652 (quoting United States v. Place, 462 U.S. 696, 707, 103 S.Ct. 2637, 77 L.Ed.2d 110 (1983)). Notably, however, Place held that a dog's detection of narcotics in luggage did not provide probable cause for a prolonged seizure of the luggage. See 462 U.S. at 707, 103 S.Ct. 2637. The Court described the dog sniff as “sui generis,” and explained, “We are aware of no other investigative procedure that is so limited both in the manner in which the information is obtained and in the content of the information revealed by the procedure.” Id. Snapchat's digital scans likewise reveal limited information about the files it flags. As already noted, the government did not replicate the digital scan. It opened the file and viewed the video, exposing considerably more information of significance to the government. See supra note 2. Opening and inspecting a digital file is more akin to a government agent opening and rummaging through a suitcase (for which a search warrant is generally required, absent exigent circumstances) than to a canine sniff or a field test that can only indicate the presence of potential contraband.

¶124 In an era of rapidly-advancing technology, including the deployment of artificial intelligence tools that collect and analyze vast amounts of data, the majority's application of the private search doctrine creates troubling precedent. This court, in holding that opening the file and viewing the video was no different than Snapchat's digital scan, sanctions greater government intrusion in reliance on private companies’ technological tools. Condoning warrantless government searches that surpass a commercial entity's previous intrusion into places in which citizens reasonably expect privacy undermines the Fourth Amendment's protections against unreasonable searches.

IV. APPLICATION OF THE EXCLUSIONARY RULE

¶125 The exclusionary rule excludes “evidence discovered during an unlawful search or seizure,” as well as “evidence discovered only because of what the police learned from the unlawful activity, also referred to as ‘fruit of the poisonous tree.’ ” State v. Van Linn, 2022 WI 16, ¶11, 401 Wis. 2d 1, 971 N.W.2d 478 (citing State v. Knapp, 2005 WI 127, ¶24, 285 Wis. 2d 86, 700 N.W.2d 899). The exclusionary rule applies to state court proceedings under the Due Process Clause of the Fourteenth Amendment. See Mapp v. Ohio, 367 U.S. 643, 81 S.Ct. 1684, 6 L.Ed.2d 1081 (1961). Since Mapp, the U.S. Supreme Court over time has curtailed the circumstances in which the exclusionary rule applies, focusing narrowly on its value in deterring police misconduct. See Richard M. Re, The Due Process Exclusionary Rule, 127 Harv. L. Rev. 1885, 1887 (2014). As we have recognized, the Court has held that the rule is properly applied only “to deter deliberate, reckless, or grossly negligent conduct, or in some circumstances recurring or systemic negligence.” See State v. Burch, 2021 WI 68, ¶17, 398 Wis. 2d 1, 961 N.W.2d 314 (quoting Herring v. United States, 555 U.S. 135, 144, 129 S.Ct. 695, 172 L.Ed.2d 496 (2009)).

A. The Good-Faith Exception

¶126 The U.S. Supreme Court has held that the exclusionary rule need not be applied to evidence obtained in violation of the Fourth Amendment when law enforcement officers relied, in objective good faith, on a judicially issued warrant or other apparent legal authority. See generally United States v. Leon, 468 U.S. 897, 104 S.Ct. 3405, 82 L.Ed.2d 677 (1984). We likewise have adopted a good-faith exception to the exclusionary rule under Article I, Section 11 of the Wisconsin Constitution. See State v. Eason, 2001 WI 98, ¶¶73–74, 245 Wis. 2d 206, 629 N.W.2d 625.

¶127 The Supreme Court has applied the good-faith exception only under circumstances showing that officers reasonably relied on then-existing legal authority in conducting a search or seizure later deemed unconstitutional. The rule was originally applied to evidence obtained in objective good-faith reliance on a judicially issued warrant. See Leon, 468 U.S. at 922, 104 S.Ct. 3405; Massachusetts v. Sheppard, 468 U.S. 981, 987–88, 104 S.Ct. 3424, 82 L.Ed.2d 737 (1984); Arizona v. Evans, 514 U.S. 1, 16, 115 S.Ct. 1185, 131 L.Ed.2d 34 (1995) (applying good-faith exception to evidence collected incident to an arrest under a quashed arrest warrant that remained active due to clerical error). This court has similarly applied the good-faith exception to unconstitutionally obtained evidence when the police reasonably relied on a facially valid search warrant. See Eason, 245 Wis. 2d 206, ¶73, 629 N.W.2d 625.⁴

¶128 The Court has also applied the exception when the government demonstrated that officers had relied, in objective good faith, on other binding legal authority, such as a statute. See Michigan v. DeFillippo, 443 U.S. 31, 38, 99 S.Ct. 2627, 61 L.Ed.2d 343 (1979) (applying good-faith exception to evidence discovered in a search incident to arrest for violating an ordinance later held to be unconstitutional); Illinois v. Krull, 480 U.S. 340, 349–50, 107 S.Ct. 1160, 94 L.Ed.2d 364 (1987) (applying good-faith exception to evidence obtained in an administrative search of vehicles in a wrecking lot pursuant to a statute later found unconstitutional).⁵

¶129 Most on point here, the Court has held that “[e]vidence obtained during a search conducted in reasonable reliance on binding precedent is not subject to the exclusionary rule.” Davis v. United States, 564 U.S. 229, 241, 131 S.Ct. 2419, 180 L.Ed.2d 285 (2011). Justice Sotomayor took care to note: “This case does not present the markedly different question whether the exclusionary rule applies when the law governing the constitutionality of a particular search is unsettled.” Id. at 250, 131 S.Ct. 2419 (Sotomayor, J., concurring) (emphasis added).

¶130 This court has similarly applied the good-faith exception to evidence obtained when officers conduct a search in reasonable reliance on clear and settled Wisconsin precedent, even if that precedent is later deemed unconstitutional by the U.S. Supreme Court. See State v. Dearborn, 2010 WI 84, ¶46, 327 Wis. 2d 252, 786 N.W.2d 97. Like Justice Sotomayor, the court cautioned that “under our holding today, the exclusionary rule is inappropriate only when the officer reasonably relies on clear and settled precedent. Our holding does not affect the vast majority of cases where neither this court nor the United States Supreme Court have spoken with specificity in a particular fact situation.” Id. (emphasis added).

¶131 Other courts have reached the same conclusion. As one district court observed, “permitting officers to rely on non-binding precedent would allow officers to pick and choose what law to follow, and would not properly serve the deterrent function of the exclusionary rule.” United States v. Robinson, 903 F. Supp. 2d 766, 782–83 (E.D. Mo. 2012) (collecting cases), aff'd, 781 F.3d 453 (8th Cir. 2015). See also United States v. Holmes, 121 F.4th 727, 735 (9th Cir. 2024) (exception does not apply to agent's warrantless view of file attached to CyberTip where “the legal landscape only made plausible the contention that [the agent's] search fell within the scope of the private-search doctrine”); Braun, 798 F. Supp. 3d at 930 (declining to apply good-faith exception to officer's warrantless view of unopened file attached to CyberTip and noting that “when the law is unsettled, officers should be encouraged to err on the side of obtaining a warrant, particularly where, as here, there is no exigency”). Put differently, crediting the government's “good faith” when it relies on favorable non-binding authority in an unsettled area of law to justify its failure to obtain a search warrant, while it ignores adverse authority, undermines the purposes of the exclusionary rule. “[W]hile an officer may reasonably rely on firm, binding precedent, the lack of binding precedent is not evidence of good faith.” Young v. State, 394 So. 3d 1174, 1179–80 (Fla. Dist. Ct. App. 2024) (emphasis added). To say otherwise “would incentivize warrantless searches under unsettled areas of law, while the Fourth Amendment requires a warrantless search to be specifically authorized by law.” Id. at 1183.

¶132 In this case, the government indisputably did not rely on settled precedent when it inspected Gasper's file without first obtaining a search warrant. Moreover, it was aware that the federal circuit courts were divided on whether a search warrant is required under similar circumstances. Compare Wilson, 13 F.4th 961, with Reddick, 900 F.3d 636, and Miller, 982 F.3d 412. Instead of taking the course of action that would ensure it was acting constitutionally—applying for a search warrant—the government instead chose to risk violating Gasper's rights. The State's policy was to open and view all of the files attached to CyberTips without a search warrant, despite knowing full well that the law was unsettled. Applying the exclusionary rule here would serve the purpose of deterring the government's deliberate choice to evade the warrant requirement. The good-faith exception should not be applied to reward the government's strategic avoidance of its obligations under the Fourth Amendment.

¶133 The U.S. Supreme Court has made clear what police must do—and what the DOJ and Sheriff's Department failed to do—before searching private data: “get a warrant.” Riley v. California, 573 U.S. 373, 403, 134 S.Ct. 2473, 189 L.Ed.2d 430 (2014); Carpenter v. United States, 585 U.S. 296, 317, 138 S.Ct. 2206, 201 L.Ed.2d 507 (2018). And in cases of doubt, this court's own precedent mandates that the government choose the course of action that avoids a constitutional violation. See Dearborn, 327 Wis. 2d 252, ¶46, 786 N.W.2d 97; see also Burch, 398 Wis. 2d 1, ¶83, 961 N.W.2d 314 (Dallet, J., concurring in part, dissenting in part) (“[B]ecause the police may encounter circumstances that are on the margins of the law regarding warrant exceptions ․ police officers are required to ‘err on the side of constitutional behavior’ and get a warrant.”) (citation modified). There was no exigency compelling the government to risk a constitutional violation. Over 60 days passed between the DOJ's initial receipt of the CyberTip and the Waukesha detective's eventual application for a warrant to search Gasper's home and devices. There was no exigency. The DOJ could have applied for and obtained a search warrant authorizing it to open and view the file. This task is a small price to pay to safeguard the rights protected by the Fourth Amendment.

B. The Evidence Obtained Under the Search Warrant

¶134 “As applied to circumstances where an application for a warrant contains both tainted and untainted evidence, the issued warrant is valid if the untainted evidence is sufficient to support a finding of probable cause to issue the warrant.” State v. Carroll, 2010 WI 8, ¶44, 322 Wis. 2d 299, 778 N.W.2d 1. “To establish probable cause to search, the evidence must indicate a fair probability that the particular place contains evidence of a crime.” Id., ¶28 (citation modified).

¶135 Thus, although I would hold that the contents of the Snapchat video were properly excluded due to the State's failure to obtain a warrant to open and view it, the State had sufficient untainted evidence to establish probable cause for the warrant to search Gasper's home and devices. The evidence seized pursuant to the search warrant, which includes all ten of the videos Gasper was charged with possessing, needs not have been suppressed.

¶136 The State argued that the results of Snapchat's digital scan—the “hash-value match”—can provide probable cause for a search warrant, even when the investigator does not view the flagged file. I agree, as have other courts. See Maher, 120 F.4th at 319 (holding that police could have relied on Google's hash-value match with known CSAM to “demonstrate probable cause to support warrants for [the government's] searches of Maher's Google accounts and residence”) (emphasis added); United States v. Cartier, 543 F.3d 442, 446 (8th Cir. 2008) (holding that hash-value match with known CSAM supported probable cause for search warrant of defendant's computer, even though no one had observed CSAM on the computer).

¶137 The CyberTip and other lawfully collected evidence (the name and birth date linked to Gasper's Snapchat account, the IP address, and his home address) would have provided probable cause for the issuance of a search warrant not only to view the video, but also to search Gasper's home and electronic devices for CSAM. It is probable that an individual who has placed a file in his ESP account has duplicate or original copies of the file on a cell phone or other device used to access the account (as explained in the affidavit, this is the case even if the files are deleted from the device). Thus, although the detailed description of the video in the search warrant affidavit was highly probative in establishing probable cause, I conclude that the probable cause standard was met even without that description. Because the evidence obtained from the lawful search pursuant to the warrant was obtained independently from the constitutional violation, it need not be excluded.

***

¶138 ESPs have many reasons for wanting to keep their platforms free of harmful and inappropriate content like CSAM, and many utilize software to monitor and prevent it. However, those efforts do not open the door to warrantless searches by the government of ESP users’ private, password-protected data. By opening and viewing Gasper's video without a search warrant, the State exceeded the bounds of the ESP's private search. The State did not do so with the virtual certainty that it would find nothing of significance in the file. Its visual examination of the video not only confirmed that the file contained CSAM, but it revealed specific images that the State described in detail in the search warrant affidavit. Because the State's decision to forego a search warrant before opening and viewing the video cannot be excused under the good-faith exception, the evidence obtained from the warrantless viewing should have been suppressed. Nevertheless, I would hold that the CSAM found during the execution of the search warrant need not be suppressed because the warrant was supported by sufficient untainted evidence.

¶139 For the foregoing reasons, I concur in part and dissent in part.

FOOTNOTES

1.   Child pornography is commonly referred to as child sexual abuse material (“CSAM”); hereinafter we use CSAM.

2.   Gasper accessed his Snapchat account exclusively from his cell phone.

3.   The CyberTipline is a website operated by NCMEC to receive and process reports of online child sexual exploitation from the public and electronic service providers (“ESPs”). The information from these reports is then shared with the appropriate law enforcement agencies for investigation and action. We refer to the report hereinafter as a “CyberTip.”

4.   MD5 is a cryptographic hash 128-bit algorithm. What is MD5? Understanding Message-Digest Algorithms, https://www.okta.com/identity-101/md5/.

5.   Gasper does not argue that Snapchat's search was anything other than private. For example, he does not argue that Snapchat was a government actor.

6.   Although we affirm the court of appeals’ ultimate judgment reversing the circuit court's grant of Gasper's motion to suppress, we do so on other grounds. Thus, we clarify that the court of appeals’ reasoning is vacated and its published opinion in this case has no precedential value.

7.   According to Snapchat's 2023 Transparency Report, Snapchat uses PhotoDNA and Google's Child Sexual Abuse Imagery Match to “identify known illegal images and videos of child sexual abuse.” Snap Inc., Transparency Report: Combating Child Sexual Exploitation & Abuse, www.values.snap.com, https://values.snap.com/privacy/transparency-h1-2023 (last updated Dec. 13, 2023).

8.   A hash value is “a string of characters obtained by processing the contents of a given computer file and assigning a sequence of numbers and letters that correspond to the file's contents.” See United States v. Reddick, 900 F.3d 636, 637 (5th Cir. 2018). Traditional hash-value scanning programs like MD5 derive a hash value based on each individual, unique piece of data in a file. Thus, if even a single pixel in an image is altered, the resulting hash value would be substantially different. A predator could therefore make an insubstantial change to a CSAM image and avoid detection. To combat this, Microsoft developed PhotoDNA, which returns the same value even if there are slight changes made to a file. See United States v. Miller, 982 F.3d 412, 418 (6th Cir. 2020).

9.   A hash match occurs when hash values are compared and are found to be identical. A hash value has been called a file's fingerprint, VIN number or DNA. See Miller, 982 F.3d at 418.

10.   Miranda v. Arizona, 384 U.S. 436, 86 S.Ct. 1602, 16 L.Ed.2d 694 (1966).

11.   All subsequent references to the Wisconsin Statutes are to the 2023—24 version unless otherwise indicated.

12.   In a footnote in its brief filed in the circuit court responding to Gasper's motion to suppress, the State engaged in a general foundational discussion of hash algorithms and their reliability, citing What is MD5? Understanding Message-Digest Algorithms, https://www.okta.com/identity-101/md5/, and Richard P. Salgado, Fourth Amendment Search and the Power of the Hash, 119 Harv. L. Rev. F. 38, 38—39 (2005). The State did not provide these articles to indicate that MD5 technology was used in the case at issue. Interestingly, the circuit court selected the okta.com article to cite in its decision.

13.   In our case, the circuit court essentially determined that Gasper had a reasonable expectation of privacy because he used a cell phone to access Snapchat, citing Riley v. California, 573 U.S. 373, 134 S.Ct. 2473, 189 L.Ed.2d 430 (2014), and Carpenter v. United States, 585 U.S. 296, 138 S.Ct. 2206, 201 L.Ed.2d 507 (2018). The circuit court also distinguished cases where an individual is found not to have a reasonable expectation of privacy in documents that were conveyed to and viewed by third parties. E.g., Smith v. Maryland, 442 U.S. 735, 742, 99 S.Ct. 2577, 61 L.Ed.2d 220 (1979) (recording of numbers dialed on a “pen register” installed by telephone company at request of law enforcement did not violate the Fourth Amendment).It should be noted that the “third party” doctrine—which holds that an individual lacks a reasonable expectation of privacy in documents willingly conveyed to third parties—is separate and distinct from the “private search” doctrine, which applies when a third party actually performs a search of an individual's secure property. Compare United States v. Miller, 425 U.S. 435, 442, 96 S.Ct. 1619, 48 L.Ed.2d 71 (1976) (third-party doctrine applied to bank receipts conveyed to financial institution), with United States v. Jacobsen, 466 U.S. 109, 104 S.Ct. 1652, 80 L.Ed.2d 85 (1984) (private party search when Federal Express employees opened suspicious package). See also Smith v. Maryland, 442 U.S. at 743–44, 99 S.Ct. 2577 (“[A] person has no legitimate expectation of privacy in information he voluntarily turns over to third parties.”).

14.   Gasper does not argue that Snapchat is acting at the direction of the government. The Fourth Amendment “protects against [private] intrusions if the private party acted as an instrument or agent of the Government.” Skinner v. Ry. Lab. Execs. Ass'n, 489 U.S. 602, 614, 109 S.Ct. 1402, 103 L.Ed.2d 639 (1989).

15.   Gasper also sought suppression of additional CSAM evidence recovered from the execution of a search warrant, on the basis that the subsequent search was the fruit of a warrantless, unconstitutional search of the Snapchat video.

16.   We note, while federal law does not require an ESP to scan for CSAM, if CSAM is found, federal law does require the ESP to notify the authorities of an “apparent violation” which contains “[CSAM].” 18 U.S.C. § 2258A(a).

17.   Jacobsen was a majority opinion. In Jacobsen, the Court also determined that the agents’ destroying cocaine to test it was a seizure that had “a de minimis impact on any protected property interest.” Jacobsen, 466 U.S. at 125, 104 S.Ct. 1652.

18.   Some courts have noted that a particular scanning system's reliability might be challenged so to affect the virtual certainty standard, but that has not been demonstrated in the case at issue. See infra note 21.

19.   Other state courts have adopted the reasoning in Miller and Reddick. See Walker v. State, 2023 Ark. App. 295, 669 S.W.3d 243, 252–55 & n.8 (Ark. Ct. App. 2023); People v. Wilson, 270 Cal. Rptr. 3d 200, 220–25 (Cal. Ct. App. 2020); Morales v. State, 274 So. 3d 1213, 1217–18 (Fla. Dist. Ct. App. 2019).

20.   Gasper's argument hinges on the fact that no person at Snapchat viewed the video. He does not argue that Snapchat's PhotoDNA scanned, opened, and inspected every communication in Gasper's account. Nor do we conclude that law enforcement would be granted access to the entirety of Gasper's Snapchat account because of Snapchat's PhotoDNA scan. The fact of this case is that Snapchat, a private actor, scanned a 16-second video, flagged it as CSAM, and turned it over to law enforcement.

21.   We note that when considering Gasper's suppression motion, the circuit court conflated reliability of MD5 scanning programs with the private search doctrine. Gasper presented no witnesses to challenge the reliability of the PhotoDNA hash-matching technology, and yet the circuit court concluded that PhotoDNA used MD5 and that MD5 is categorically unreliable. The circuit court based its decision on the risk of “collision” and the unreliability of MD5 algorithms. The circuit court appears to have concluded that the MD5 algorithm played a role in the PhotoDNA scan. It stated, “PhotoDNA assigned Gasper's video a hash value that starts with ‘MD5.’ ․ If the ‘MD5’ is unreliable, it will create a ‘collision.’ ” The record instead reflects that PhotoDNA is reliable and there is no evidence that MD5 was used. The detective did testify that an MD5 hash value was provided by NCMEC to assist law enforcement in future investigations. The detective stated he did not use MD5. The record does not support the circuit court's conclusion that PhotoDNA assigned the MD5 hash value, that MD5 was even used, or that there was MD5 collision. There is nothing in the record which indicates that PhotoDNA uses MD5, that Snapchat used MD5, or that anyone used MD5. Here, the CyberTip itself stated that PhotoDNA identified the reported file as known CSAM. The record reflects that the identifier operates like a “serial number” for the file. PhotoDNA is used to ensure there is no alteration that will avoid detection and instead PhotoDNA divides and identifies the scanned file compared to known CSAM files into tiny pieces, deriving a hash value for each piece and then comparing the pieces. Consequently, on this record, PhotoDNA reliably flagged the video in comparison to known CSAM files. The circuit court's factual findings regarding MD5 were clearly erroneous.

22.   This CyberTip is somewhat like a tip from an identified citizen informant. The fact that federal law requires that ESPs like Snapchat report “apparent [CSAM]” as the ESPs “becom[e] aware,” has caused some courts to conclude that this requirement actually heightens the reliability of the tip. State v. Silverstein, 2017 WI App 64, ¶19, 378 Wis. 2d 42, 902 N.W.2d 550. Silverstein aptly compares a CyberTip to that of an identified citizen informant.

23.   Nor does our private party search analysis need to turn on whether Gasper violated Snapchat's terms of service.

24.   No law enforcement officer accessed Gasper's Snapchat account or cell phone without a warrant. DOJ issued an administrative warrant and law enforcement later obtained a search warrant after it also confirmed the video contained CSAM.

1.   A different analysis would be needed had Snapchat provided complete access to, or all videos from, Gasper's account or other downloads which were not flagged as CSAM. But it did not. A different analysis might occur if the algorithm deployed was proven unreliable. But it was not.

2.   “Secondary trauma affects people who witness traumatic events ․ It happens when people are exposed to another person's traumatic event.” Kendall-Tackett, Kathleen, Psychological Trauma Theory, Research, Practice & Policy, Editorial, 15 Journal of the American Psychological Association, No. S2, S201-S202 (2023).

1.   For ease of discussion, I use the phrase “terms of service” throughout this opinion to refer collectively to three separate Snapchat policies: its terms of service, community guidelines, and “Sexual Content Community Guidelines Explainer Series.” Each of these documents are available from Snapchat, and users must agree to their terms in order to join the platform.

2.   At times, the court of appeals described its opinion as deciding “whether Gasper had a reasonable expectation of privacy in the video in his Snapchat account.” Gasper, 414 Wis. 2d 532, ¶15, 16 N.W.3d 279 (emphasis added); see also id., ¶12 (“Gasper's Snapchat account [is] the relevant ‘area’ that was searched.” (quoting another source)). This framing is incorrect, however, since no government official ever searched Gasper's account; the Fourth Amendment “search,” if any occurred (and, as Justice Crawford explains in her separate writing, a search did occur), happened when law enforcement viewed for the first time the video attached to the tip by Snapchat and NCMEC. The issue is thus whether Gasper had a reasonable expectation of privacy in the video. Despite the court of appeals’ mistaken framing, I discuss its reasoning further, since it would apply equally to the video alone. The video, after all, was subject to the same Snapchat policies as Gasper's account, policies the court of appeals said eliminated any expectation of privacy Gasper otherwise had. See id., ¶¶21–28.

3.   In discussing why Gasper lacked a reasonable expectation of privacy in the video, the court of appeals noted that uploading the video to his account “was obviously unlawful.” See Gasper, 414 Wis. 2d 532, ¶22, 16 N.W.3d 279. In context, the court of appeals was describing why Gasper's conduct violated Snapchat's terms of service. It could not have meant—as the majority erroneously suggests—that Gasper lacked a reasonable expectation of privacy simply because his conduct was unlawful. See majority op., ¶4. The United States Supreme Court has made clear that individuals may have reasonable expectations of privacy even while engaging in criminal activity. See, e.g., Minnesota v. Olson, 495 U.S. 91, 96–97, 110 S.Ct. 1684, 109 L.Ed.2d 85 (1990); Payton v. New York, 445 U.S. 573, 583–603, 100 S.Ct. 1371, 63 L.Ed.2d 639 (1980); Katz v. United States, 389 U.S. 347, 348–59, 88 S.Ct. 507, 19 L.Ed.2d 576 (1967). As Justice Ginsburg aptly explained, “[i]f the illegality of the activity made constitutional an otherwise unconstitutional search, such Fourth Amendment protection, reserved for the innocent only, would have little force in regulating police behavior toward either the innocent or the guilty.” Minnesota v. Carter, 525 U.S. 83, 110, 119 S.Ct. 469, 142 L.Ed.2d 373 (1998) (Ginsburg, J., dissenting).

4.   See, e.g., United States v. Colbert, No. 23-cr-40019-TC-1, 2024 WL 2091995, at *8–9 (D. Kan. May 9, 2024) (holding that Snapchat's terms of service eliminated a user's expectation of privacy in his account); United States v. Sporn, No. 21-10016-EFM, 2022 WL 656165, at *9–10 (D. Kan. Mar. 4, 2022) (concluding that a violation of Twitter's terms of service meant a user lacked a reasonable expectation of privacy in his account); United States v. Bohannon, 506 F. Supp. 3d 907, 915 (N.D. Cal. 2020) (stating that by agreeing to the terms of service, a user consented to a search of his Microsoft OneDrive account); but see United States v. Warshak, 631 F.3d 266, 286 (6th Cir. 2010) (explaining that although terms of service “might, in some cases, be sweeping enough to defeat a reasonable expectation of privacy in the contents of an email account,” the AOL terms at issue in the case did not do so); United States v. Irving, 347 F. Supp. 3d 615, 623 (D. Kan. 2018) (determining that Facebook's terms of service did not eliminate a user's expectation of privacy in his account's contents).

5.   United States v. Ackerman, 831 F.3d 1292, 1294 (10th Cir. 2016) (citing Richard P. Salgado, Fourth Amendment Search and the Power of the Hash, 119 Harv. L. Rev. F. 38, 38–40 (2005)).

6.   This was equally true in the other case Justice Hagedorn's concurrence cites, United States v. Simpson, 904 F.2d 607, 610 (11th Cir. 1990). There, a private party opened a box containing magazines and videotapes, viewed the tapes, and turned them over to law enforcement. See id. at 608–09. Because “[t]he box's contents had already been examined, their illicit character had been determined, and they were open for viewing by the time the Assistant United States Attorney and the F.B.I. Agent arrived on the scene,” law enforcement's actions in that case clearly fell within the scope of the private-search doctrine. Id. at 610. Unlike in Simpson, no private party viewed the video at issue in this case before the DOJ analyst.

7.   See, e.g., Maher, 120 F.4th at 315–16; United States v. Wilson, 13 F.4th 961, 978–79 (9th Cir. 2021); United States v. Miller, 982 F.3d 412, 429 (6th Cir. 2020); Ackerman, 831 F.3d at 1305–06; but see United States v. Reddick, 900 F.3d 636, 639 (5th Cir. 2018)

8.   We could, of course, impose additional requirements on the application of the good-faith exception under the Wisconsin Constitution. See State v. Eason, 2001 WI 98, ¶74, 245 Wis. 2d 206, 629 N.W.2d 625. Gasper did not make an argument under the Wisconsin Constitution for limiting the good-faith exception in this situation, however, and I therefore apply only the standards applicable under the United States Constitution.

9.   Subsequently, the Second Circuit's decision in Maher joined Wilson’s side of this split of authority. See generally Maher, 120 F.4th 297. Nonetheless, that case was not decided until after this case was on appeal, and thus is not relevant to evaluating whether to apply the exclusionary rule. Cf. Davis, 564 U.S. at 240–41, 131 S.Ct. 2419.

1.   In this case, the government's viewing of the file confirmed what it suspected: that the defendant possessed CSAM. But it is not hard to imagine scenarios in which the government's viewing of a video flagged as CSAM by a digital scan would uncover additional criminal activity, such as when a video depicts the defendant engaging in sexual activity with a child or depicts a child known to the defendant. See, e.g., United States v. Runyan, 275 F.3d 449 (5th Cir. 2001) (officers’ searches of photos on defendant's laptop and disks revealed evidence that he sexually exploited a child by producing CSAM); see Wis. Stat. §§ 948.02 (sexual assault of a child), 948.05 (sexual exploitation of a child, including producing CSAM), 948.051 (trafficking of a child), 948.07 (child enticement). Gasper himself was charged with nine counts of sexual exploitation of a child for distributing videos later found on his cell phone pursuant to the search warrant.

2.   A video file, in basic terms, is a string of digital code that can be processed by a computer or other device, using a compatible program, to display a video. The contents of the file cannot be observed with the naked eye. See generally Catherine Guthrie & Brittan Mitchell, The Swinton Six: The Impact of State v. Swinton on the Authentication of Digital Images, 36 Stetson L. Rev. 661, 662 (2007).

3.   Justice Hagedorn's hypothetical about a woman conducting a keyword search on her spouse's emails and handing printed copies of the emails to the police is readily distinguishable on this point. Unlike a digital file, the printed emails place the incriminating evidence in plain view. Law enforcement officers, upon being handed such emails, need not “avert their eyes.” Coolidge v. New Hampshire, 403 U.S. 443, 489, 91 S.Ct. 2022, 29 L.Ed.2d 564 (1971). Justice Dallet makes a similar point in her concurrence. See Justice Dallet's concurrence, ¶83.

4.   Notably, this court held that, under Article I, Section 11 of the Wisconsin Constitution, additional safeguards must be present for the good-faith exception to apply in the context of a search warrant:We hold that where police officers act in objectively reasonable reliance upon the warrant, which had been issued by a detached and neutral magistrate, a good faith exception to the exclusionary rule applies. We further hold that in order for a good faith exception to apply, the burden is upon the State to show that the process used in obtaining the search warrant included a significant investigation and a review by either a police officer trained and knowledgeable in the requirements of probable cause and reasonable suspicion, or a knowledgeable government attorney.State v. Eason, 2001 WI 98, ¶74, 245 Wis. 2d 206, 629 N.W.2d 625. I note that Gasper does not make any argument here urging this court to limit the scope of the good-faith exception under Article I, Section 11 of the Wisconsin Constitution.

5.   On the other hand, the Supreme Court has repeatedly declined to apply the good-faith exception to searches conducted under the authority of a statute purporting to authorize a search without probable cause or a valid warrant. See Michigan v. DeFillippo, 443 U.S. 31, 39, 99 S.Ct. 2627, 61 L.Ed.2d 343 (1979) (collecting cases).

Annette Kingsland Ziegler, J.