Trulock v. Freeh

MICHAEL, Circuit Judge,

concurring in part and dissenting in part:

I dissent from part III.C. of the majority’s opinion, but otherwise concur. The owner of password-protected computer files has a clear expectation of privacy in those files that is protected by the Fourth Amendment. Another person who does not know the passwords has no authority to consent to a search of these private files because he lacks the “joint access or control” required by United States v. Matlock, 415 U.S. 164 , 171 n. 7, 94 S.Ct. 988 , 39 L.Ed.2d 242 (1974). The plaintiffs, Notra Trulock and Linda Conrad, both used Conrad’s computer, but each maintained personal files that were protected by passwords. Conrad told the defendant-FBI agents that she did not know either the passwords for Trulock’s files or the contents of those files. As a result, Conrad’s general consent to a search of her computer could not authorize the FBI’s warrantless search of Trulock’s password-protected files. This should have been abundantly clear to any reasonable law enforcement officer operating in the year 2000. I therefore respectfully dissent from the majority’s decision to affirm the grant of qualified immunity to the FBI officials on the search of Tru-lock’s password-protected computer files. On a separate point, I agree with the majority that the officials are entitled to qualified immunity on their warrantless search of Conrad’s house, but I write independently to explain why I reach that conclusion.

I.

The majority holds that Conrad lacked the authority to consent to a search of Trulock’s password-protected computer files. Ante at 406. I agree. I also agree with the majority’s conclusion, see id., that *407 Trulock’s computer files are analogous to the locked footlocker in United States v. Block, 590 F.2d 535, 540-42 (4th Cir.1978) (holding that a mother’s consent to the search of her son’s room did not extend to his locked footlocker). I respectfully disagree, however, with the majority’s view that the defendants are entitled to qualified immunity because there was no clearly established law saying that one co-user’s consent to search a computer does not extend to the password-protected files of another co-user when the consenting co-user does not know the other’s passwords. I would reject the defendants’ qualified immunity defense because the unlawfulness of searching Trulock’s password-protected files was readily apparent in light of the principles established in Matlock and reiterated in Block .

Qualified immunity shields a government official from civil liability so long as his conduct “does not violate clearly established statutory or constitutional rights of which a reasonable person would have known.” Harlow v. Fitzgerald, 457 U.S. 800, 818 , 102 S.Ct. 2727 , 73 L.Ed.2d 396 (1982). In deciding whether a right is clearly established for qualified immunity purposes, the question is not whether the general right (here, the right to be free from unreasonable searches) is clearly established. Rather, the question is whether the right is clear in relation to the specific conduct being challenged. See Wiley v. Doory, 14 F.3d 993, 995 (4th Cir.1994). In other words, “[t]he contours of the right must be sufficiently clear that a reasonable official would understand that what he is doing violates that right.” Anderson v. Creighton, 483 U.S. 635, 640 , 107 S.Ct. 3034 , 97 L.Ed.2d 523 (1987). This does not mean, however, that “the very action in question [must have] previously been held unlawful.” Id. Liability will attach if the unlawfulness of the conduct would be “apparent” to a reasonable officer “in the light of pre-existing law.” Wilson v. Layne, 526 U.S. 603, 615 , 119 S.Ct. 1692 , 143 L.Ed.2d 818 (1999). For the class of “clearly established” rights “includes not only already specifically adjudicated rights, but those manifestly included within more general applications of the core constitutional principle invoked.” Pritchett v. Alford, 973 F.2d 307, 314 (4th Cir. 1992) (emphasis added).

The central question here is whether in the light of pre-existing law it would have been apparent to a reasonable FBI agent that Conrad’s general consent to search the computer she shared with Trulock did not authorize the search of Trulock’s password-protected files stored in that computer. In answering this question, we look to Supreme Court cases, “ ‘cases of controlling authority in [this] jurisdiction,’ [and] the ‘consensus of cases of persuasive authority’ from other jurisdictions” as sources of clearly established law. Amaechi v. West, 237 F.3d 356, 363 (4th Cir. 2001) (quoting Wilson, 526 U.S. at 617 , 119 S.Ct. 1692 ). No court has decided a case involving third-party consent to the search of password-protected computer files. Nevertheless, we have clearly established law that is applicable: it comes from Matlock and Block .

A warrantless search can sometimes be authorized by a third party who is not the target of the search. Matlock established that third-party consent is valid only when the third party and the target have “common authority” over the area or item sought to be searched. Matlock, 415 U.S. at 171 , 94 S.Ct. 988 . Common authority, the Supreme Court explained, rests on “mutual use of the property by persons generally having joint access or control for most purposes.” Id. at 171 n. 7, 94 S.Ct. 988 . When common authority exists, the target of a search has “assumed the risk” that another person with authority over a *408 shared area or item might consent to a search. Id. The principle that valid third-party consent requires common authority should be sufficient to defeat the defendants’ claims of qualified immunity in this case. Specifically, it should have been obvious to law enforcement officials operating in the year 2000 that common authority over password-protected computer files requires knowledge of the passwords. 1 Even so, the government argues that because Conrad had common authority over the computer she shared with Trulock, the defendant — FBI agents reasonably failed to understand that her general consent to search the computer did not authorize the search of all files stored in the computer. The government’s argument fails because a reasonable officer who understood our decision in Block would have known that the search of Truloek’s private files was unlawful.

In Block we applied Matlock in deciding whether a third party’s consent to the search of a general area over which she has common authority validates the search of every item within that area. The mother in Block had general access to the room in which her defendant son’s footlocker was located, and she signed a written consent form authorizing a “complete search” of her son’s room. Block, 590 F.2d at 537 n. 1. Nevertheless, we held that the mother’s consent did not authorize the search of her son’s footlocker. We emphasized that authority to consent to the search of a general area “cannot be thought automatically to extend to the interiors of every discrete enclosed space capable of search within the area.” Id. at 541 . An enclosed space or distinct item requires independent consent for a search when the circumstances indicate that the person targeted has “a discrete expectation of privacy with respect to the particular [space or item].” Id. at 541 n. 8. Privacy expectations are signaled, for example, when the space or item is secured or “is commonly used for pre-serv-ing privacy.” Id. This means, in other words, that a third party’s common authority ends where the target’s discrete expectation of privacy begins. In sum, Block announced the general principle that when a third party and the target of a search have common authority over a general area, the third party’s consent to a search of the general area does not authorize the search of a specific item within that area if the circumstances indicate that the target has a discrete expectation of privacy with respect to that item. This principle dictates the result in this case.

The majority readily agrees that “Tru-lock’s password-protected files are analogous to the locked footlocker inside the bedroom” in Block and that Trulock has therefore “alleged a violation of his Fourth Amendment rights.” Ante at 403. That conclusion is unassailable because the factual parallels between this case and Block are striking. The mother in Block had common authority over her son’s bedroom, just as Conrad had common authority over the computer she shared with Trulock. The mother gave consent to search the bedroom, just as Conrad gave consent to search the computer. The mother told the police that the footlocker belonged to her son, that he kept it locked, and that she did not have the key. Block, 590 F.2d at 538 . Conrad told the FBI agents that she did not know what information Trulock kept in his computer files and that she *409 could not access those files because she did not know the passwords. Just as the mother’s consent to the search of her son’s bedroom did not extend to his locked footlocker inside that room, Conrad’s consent to the search of her computer did not extend to Trulock’s “locked” files inside that computer. Indeed, the only notable difference between the two cases is that Block involved a locked footlocker and this case involves password-protected computer files. For the majority, however, the immunity decision turns on this one difference. The majority gives qualified immunity to the defendants because of its reluctance to “recognize a retroactive right based on cases involving footlockers and other dissimilar objects.” Ante at 404. In essence, the majority is hesitant to hold the FBI agents responsible for applying Block’s clearly established legal principle in a different factual context. The agents, the majority believes, could not be expected to understand that the expectations of privacy signaled by a locked footlocker and a password-protected computer file are essentially the same.

While it is true that knowing a legal principle and knowing whether to apply it in a particular circumstance are two different things, see Lappe v. Loejfelholz, 815 F.2d 1173 , 1180 n. 7 (8th Cir.1987), qualified immunity was never intended to relieve government officials from the responsibility of applying familiar legal principles to new situations. To say otherwise would ignore the Supreme Court’s warning that liability under § 1983 (and Bivens) does not require “the very action in question [to have] previously been held unlawful.” Wilson, 526 U.S. at 615 , 119 S.Ct. 1692 . Whatever the physical differences between locked footlockers and password-protected computer files, the question here must be whether a reasonable officer would believe that there is a legal difference for Fourth Amendment purposes. In other words, is there any reason why a reasonable FBI agent fully apprised of the principles in Block would believe that he could lawfully search Trulock’s password-protected files on the basis of Conrad’s general consent to search the computer? If there is no such reason, the unlawfulness of the agents’ conduct in this case is “apparent,” Wilson, 526 U.S. at 615 , 119 S.Ct. 1692 , and qualified immunity does not apply. Cf. Lassiter v. Alabama A & M University, 28 F.3d 1146, 1150 (11th Cir.1994) (stating that qualified immunity is lost when pre-existing law “dictate[s]” or “compel[s]” the conclusion that a defendant’s conduct violates federal rights).

Any reasonable officer should have recognized that the privacy expectations attaching to a password-protected computer file are essentially the same as those attaching to a locked footlocker. A computer file is a repository for information and images in electronic form, just as a footlocker is a repository for more tangible items such as papers and other personal effects. Once password protection attaches to a computer file, that protection is the electronic equivalent of the lock on a footlocker containing items that are intended to remain private. The password is an electronic key. While the medium for ensuring privacy is different, the result — a clear signal that privacy is expected against all those who lack the key (or the password)- — is the same. There is simply no reason why a reasonable officer who understood that a locked footlocker signals a discrete expectation of privacy would believe that a password-protected computer file does not. The physical differences between the two repositories have no legal significance.

This conclusion is not undercut by the majority’s observation that the law of computers is “fast evolving.” Ante at 403. In fact, the case law supports my point that *410 the differences between computer files and physical repositories of personal information and effects are legally insignificant. Courts have not hesitated to apply established Fourth Amendment principles to computers and computer files, often drawing analogies between computers and physical storage units such as file cabinets and closed containers. See, e.g., In re Grand Jury Subpoena Duces Tecum, 846 F.Supp. 11, 12-18 (S.D.N.Y.1994) (analogizing computer hard drives and floppy disks that contained electronic documents to file cabinets that contained paper documents in deciding that subpoena for computer-accessible data was unreasonably broad); United States v. Chan, 830 F.Supp. 531, 534-35 (N.D.Cal.1993) (holding that “[t]he expectation of privacy in an electronic repository for personal data is ... analogous to that in a personal address book or other repository for such information .... [A]n individual has the same expectation of privacy in a pager, computer or other electronic data storage and retrieval device as a closed container ....”) (internal quotation and citation omitted); United States v. David, 756 F.Supp. 1385, 1390 (D.Nev.1991) (recognizing that a computer memo book “is in distinguishable from any other closed container, and is entitled to the same Fourth Amendment protection”). Our circuit has also drawn analogies between computer files and physical repositories of personal information and effects, such as lockers. See United States v. Simons, 206 F.3d 392, 398 (4th Cir.2000) (citing American Postal Workers Union v. United States Postal Serv., 871 F.2d 556 , 560 (6th Cir.1989), to compare employee lockers subject to random inspection under employer policy with computer files subject to “appropriate” inspection under employer policy allowing monitoring of employee Internet use). 2 *411 Thus, neither case law nor common sense suggests any reason for thinking that the principles in Block do not dictate the result in this case.

While the majority bases its grant of qualified immunity primarily on the factual differences between this case and Block , it also suggests that there is no clearly established law governing the search of Tru-lock’s protected files because at least one district court opinion from another circuit has upheld the search of a shared computer based on third-party consent. Ante at 403 (citing United States v. Smith, 27 F.Supp.2d 1111 (C.D.Ill.1998)). Smith , however, is consistent with Block , and its reasoning actually supports the conclusion that Conrad’s consent to search the computer did not encompass Trulock’s password-protected files.

In Smith the court upheld a search of the defendant’s computer files based on third-party consent, but the facts were significantly different from those presented here. There, the defendant’s housemate consented to a search of the defendant’s computer, which was located in an alcove in the housemate’s bedroom. The court found that the housemate had the necessary joint control and access to the computer and its surrounding area because the. computer was accessible to all members of the household, it had been used by the housemate’s daughter, and the defendant had tried to teach the housemate to use it. Smith, 27 F.Supp.2d at 1115-16 . Although there was some factual dispute about whether the defendant had used passwords to protect computer files containing images of child pornography, the court found that these files were not password protected. Id. at 1116 (“[I]t is important to note that none of the officers who searched the computer found passwords on the computer. This belies Defendant’s claim of exclusive and possessory control and indicates that [the housemate] could consent to the search of the home and computer and that the consent extended to the computer area and the computer itself.”). At most, then, Smith stands for the proposition that a third party with shared access to a computer may consent to the search of all the files on the computer that are not protected by individualized passwords. Indeed, the court’s conclusion that a lack of password protection discredits claims of exclusive possession and control suggests that the presence of such protection would establish exclusive possession and control, thereby placing the password-protected files outside the scope of valid third-party consent. As the majority recognizes, Smith held only that “a third party may consent to the search of a shared computer when the third party has complete access to the computer.” Ante at 403. Certainly Conrad had general access to the computer, and certainly a reasonable officer would have believed that Conrad had the authority to consent to a search of all of the commonly accessible files on the computer. But access to a computer need not — and here it did not-extend to each and every file on that computer. A reasonable officer aware of the principles in Block would not have thought otherwise.

I would hold, therefore, that the search of Trulock’s password-protected files violated clearly established law because the unconstitutionality of the search was readily apparent in light of the core principles applied in Matlock and Block . This position is supported by the government’s own conclusions about how Fourth Amendment principles apply to computer technology. In a manual designed to educate federal agents about the law governing searches *412 and seizures of computers, the Department of Justice (DOJ) explicitly acknowledges that “it appears likely that encryption and password-protection would in most cases indicate the absence of common authority to consent to a search among co-users who do not know the password or possess the encryption key.” Searching and Seizing Computers and Obtaining Electronic Evidence in Criminal Investigations, at p. 14 (2001), available at www.cyber-crime.gov/searchmanual.pdf. 3 It is especially striking that the DOJ based its conclusion on an analysis of Block and Smith, see id., the very authorities relied upon by the majority to assert that there was no clearly established law indicating that Conrad had no authority to consent to the search of Trulock’s password-protected files. In effect, the government now invites this court to find that the law governing third-party consent to computer searches is uncertain even though it has shown itself quite capable of correctly resolving the question presented in this case. I would decline this invitation.

Qualified immunity is intended “to protect those officers who reasonably believe that their actions do not violate federal law,” Doe v. Broderick, 225 F.3d 440, 453 (4th Cir.2000), but it should not function to give officers “one free violation” of constitutional rights every time they are asked to apply a well-established principle to a new set of facts, Wilson, 526 U.S. at 625 , 119 S.Ct. 1692 (Stevens, J., concurring in part and dissenting in part). The defendants in this case should have known that they had no right to search Trulock’s password-protected computer files, and thus they should not be given qualified immunity-

II.

In part III.B. of its opinion the majority concludes that although Conrad did not voluntarily consent to the search of her house, the defendants are entitled to qualified immunity from any liability for that search. While I agree with the majority’s conclusion, I write separately because my reasons for granting the defendants qualified immunity on the house search may differ from the majority’s.

Although Conrad signed a written consent form authorizing the FBI to search her house, she alleges that this consent was involuntary because it was prompted by her belief that the FBI already had a search warrant and that the FBI would break down her front door and search the house in the presence of the media and local police if she refused to cooperate. Although the voluntariness of consent for Fourth Amendment purposes is “a question of fact to be determined from the totality of all the circumstances,” Schneckloth v. Bustamonte, 412 U.S. 218, 227 , 93 S.Ct. 2041 , 36 L.Ed.2d 854 (1973), one factor that nearly always invalidates consent is an assertion by law enforcement officers that they have the authority to search with or without consent. See, e.g., Bumper v. North Carolina, 391 U.S. 543, 550 , 88 S.Ct. 1788 , 20 L.Ed.2d 797 (1968) (holding that consent was invalid when given after police officers claimed authority to search home under a warrant); United States v. Lattimore, 87 F.3d 647, 652 *413 (4th Cir.1996) (en banc) (stating that police officer’s assertion that he would “ ‘call a drug dog’ ” to search suspect’s car if suspect refused consent “would raise serious questions concerning the voluntariness of his consent”); Orhorhaghe v. I.N.S., 38 F.3d 488, 500 (9th Cir.1994) (stating that “[i]t is well established that there can be no effective consent to a search or seizure if that consent follows a law enforcement officer’s assertion of an independent right to engage in such conduct.”). Consent is also involuntary when officers threaten a person with adverse consequences if she refuses to consent to a search. See State v. Davis, 261 Ga. 225 , 404 S.E.2d 100, 100-01 (1991) (affirming trial court’s ruling that consent was involuntary when police told defendant’s mother they would break down the door if she refused to cooperate); Reyes v. Edmunds, 472 F.Supp. 1218, 1227-28 (D.Minn.1979) (holding that consent was involuntary when welfare recipient was told that her benefits would be terminated if she refused to consent to the search of her home). Here, Conrad alleges that the FBI, acting through Lawrence Sanchez (Conrad’s supervisor at the Department of Energy), claimed to have authority to search her house and threatened her with property damage and public humiliation if she refused to cooperate. Under these circumstances, her consent was not voluntary.

The government claims that there are two reasons why this case is not controlled by Bumper , in which the Supreme Court held that consent is involuntary when given after “the official conducting the search has asserted that he possesses a warrant.” Bumper, 391 U.S. at 548 , 88 S.Ct. 1788 . First, it argues that this case is distinguishable from Bumper because the complaint fails to allege that the FBI agents who conducted the search claimed they had a warrant or knew that Sanchez had told Conrad that they had a warrant. Second, the government argues that because Sanchez is neither a defendant nor a law enforcement official, what he allegedly said to Conrad does not bear on whether the defendants violated her constitutional rights. Like the majority, I refuse to read Bumper so narrowly. Bumper stands for the proposition that consent cannot be voluntary when the government has led the person consenting to “erroneously] belie[ve] that [s]he cannot protect [her] privacy by refusing to give consent.” 3 Wayne R. LaFave, Search and Seizure, § 8.2(c) at 652 (3rd ed.1996). If, as the complaint alleges, Sanchez was acting on behalf of the FBI, then the government led Conrad to believe that her only choice was between losing her privacy quietly and losing it in the glare of the media spotlight. That the-government conveyed this message through Sanchez rather than through the FBI agents conducting the search does not change the result under Bumper .

The majority and I may differ, however, in our reasons for concluding that the defendants are entitled to qualified immunity on the house search. Qualified immunity protects government officials who make reasonable mistakes of fact as well as those who make reasonable mistakes about what the law requires in a particular situation. Karnes v. Skrutski, 62 F.3d 485, 498 (3rd Cir.1995) (stating that qualified immunity protects those who make “ ‘mere mistakes in judgment, whether the mistake is one of fact or one of law' ” (quoting Butz v. Economou, 438 U.S. 478, 507 , 98 S.Ct. 2894 , 57 L.Ed.2d 895 (1978))). As my previous discussion of the search of Tru-lock’s computer files illustrates, qualified immunity analysis usually turns on whether the illegality of the defendant’s conduct was apparent in the light of clearly established law. Here, however, I believe the defendants are entitled to qualified immunity only because it is undisputed that they *414 did not know all of the relevant facts. Because the complaint fails to allege either that the defendants told Conrad that they had a warrant or that they knew Sanchez had told Conrad that they had a warrant, we must assume that the defendants were unaware of Conrad’s belief that they would search her house with or without her consent. Without any awareness of what Sanchez had said to Conrad, the defendants could have reasonably believed that Conrad’s written consent was valid. The defendants are therefore entitled to qualified immunity. On my analysis, then, the crucial factor in explaining why the defendants should receive qualified immunity is that they made a reasonable mistake of fact about what Conrad believed. Because the defendants did not know what Sanchez had told Conrad, they reasonably failed to recognize that Conrad believed that she could no longer protect her privacy by refusing to consent to the search of her house.

In explaining its qualified immunity holding, the majority also emphasizes the defendants’ lack of knowledge of the conversation between Sanchez and Conrad. Thus, the majority and I may agree that qualified immunity is justified only because the defendants made a reasonable mistake of fact. I wish to be explicit on the point, however, because I could not accept the proposition that the defendants in this case made a reasonable mistake of law. Specifically, I would refuse to grant the defendants qualified immunity if the complaint had alleged that any of the defendants had personally directed Sanchez to threaten Conrad or that the defendants knew that Sanchez had conveyed threats to Conrad at the behest of someone in the FBI. On those facts, the defendants’ only argument for qualified immunity would have been that the invalidity of Conrad’s consent was not readily apparent in light of the factual distinctions between this case and Bumper . For example, the government might have argued that this case differs from Bumper because there the officers who claimed to possess a warrant also conducted the search, whereas here Sanchez claimed that the FBI had a warrant but the agents who conducted the search did not make that claim. I would reject such arguments for reasons similar to those given in part I above. Bumper clearly establishes that there can be no valid consent when the government has led a person to believe that her consent is irrelevant, and there is no reason why a reasonable officer would think that the factual differences between Bumper and this case are legally significant.

III.

In sum, I agree with the majority’s disposition of this case, except that I respectfully dissent from its decision to grant the defendants qualified immunity on Tru-lock’s claim that the warrantless search of his password-protected computer files violated his Fourth Amendment rights. I would therefore reverse the district court’s order granting the defendants’ motion to dismiss that claim.

. Knowledge of the passwords is necessary, but not sufficient, to establish common authority over password-protected files. The third party must also have "joint access" to the files "for most purposes.” Matlock, 415 U.S. at 171 n. 7, 94 S.Ct. 988 . If Conrad had known Tralock's passwords and had enjoyed general access to his files, this would be a different case.

. These analogies have limitations, of course. For example, the Tenth Circuit rejected an argument based on the file cabinet analogy in deciding that a detective exceeded the scope of a search warrant when he opened certain of the defendant's computer files. See United States v. Carey, 172 F.3d 1268 (10th Cir.1999). In Carey the government argued that an officer with a warrant to search a file cabinet for files containing certain information can open every drawer of the file cabinet, even when the labels on the file drawers suggest that none of the files within that drawer fall within the scope of the warrant. Opening every drawer, the government insisted, is the only way to be sure that the labels on the file drawer are accurate. The government then argued by analogy that an officer executing a warrant to search files on a computer for specified information can also open all of the computer's files, including those files whose names suggest that they contain no information within the scope of the warrant. See id. at 1274-75 . The court held that the file cabinet analogy does not extend this far because in the case of a computer, officers may use key word searches and similar techniques to identify which files fall within the scope of a warrant without the need to open all of the files in the computer. See id. at 1275-76 ; see also Raphael Winick, Searches and Seizures of Computers and Computer Data, 8 Harv. J.L. & Tech. 75 , 103-11 (1994) (discussing the limitations of the closed container analogy and recommending that courts adopt a version of the “intermingled documents” rule adopted in United States v. Tamura, 694 F.2d 591, 595-96 (9th Cir.1982), to govern computer searches). The Carey court's limitation on the file cabinet analogy makes sense because the court relied on a feature of computer files unique to the electronic medium, namely, their amenability to key word searches and similar techniques. The detective's failure to use these techniques to limit the scope of his computer search made the search unreasonable. See Carey, 172 F.3d at 1276 . Thus, the physical differences between computer files and file cabinets made a legal difference in Carey , and the court properly warned the uncritical acceptance of the file cabinet analogy could lead courts to sanction indiscriminate searches of computer files. Nevertheless, no court to my knowledge has suggested that the differences between computer files and other repositories for personal information raise difficult problems in deciding whether a given repository signals a "discrete *411 expectation of privacy.” Block, 590 F.2d at 541 n. 8.

. On this point, the contrast with Wilson v. Layne is instructive. There, the Supreme Court buttressed its finding of qualified immunity by stating that the police reasonably relied on a U.S. Marshals Service policy governing media ride-alongs which clearly contemplated that media members might accompany police into private homes. See Wilson, 526 U.S. at 617 , 119 S.Ct. 1692 . In contrast, the agents' conduct here contravened the DOJ's own understanding of the legal norms governing third-party consent to computer searches.