Free Speech Coalition, Inc. v. Attorney General United States

PRECEDENTIAL UNITED STATES COURT OF APPEALS FOR THE THIRD CIRCUIT _____________ No. 13-3681 _____________ FREE SPEECH COALITION, INC.; AMERICAN SOCIETY OF MEDIA PHOTOGRAPHERS, INC.; THOMAS HYMES; TOWNSEND ENTERPRISES, INC., d/b/a SINCLAIR INSTITUTE; BARBARA ALPER; CAROL QUEEN; BARBARA NITKE; DAVID STEINBERG; MARIE L. LEVINE, a/k/a NINA HARTLEY; DAVE LEVINGSTON; BETTY DODSON; CARLIN ROSS, Appellants v. ATTORNEY GENERAL UNITED STATES OF AMERICA _____________ On Appeal from the United States District Court for the Eastern District of Pennsylvania District Court No. 2-09-cv-04607 District Judge: The Honorable Michael M. Baylson Argued December 9, 2014 Before: RENDELL, SMITH, and SCIRICA, Circuit Judges (Opinion Filed: May 14, 2015) Lorraine R. Baumgardner, Esq. J. Michael Murray, Esq. [ARGUED] Berkman, Gordon, Murray & DeVan 55 Public Square Suite 2200 Cleveland, OH 44113 Kevin E. Raphael, Esq. J. Peter Shindel, Jr., Esq. Pietragallo, Gordon, Alfano, Bosick & Raspanti 1818 Market Street Suite 3402 Philadelphia, PA 19103 Counsel for Appellant Hector Bladuell, Esq. James J. Schwartz, Esq. Nathan M. Swinton, Esq. Kathryn Wyer, Esq. United States Department of Justice Civil Division, Federal Programs Branch 20 Massachusetts Avenue, N.W. Room 7130 Washington, DC 20530 Scott R. McIntosh, Esq. United States Department of Justice 2 Civil Division Room 7259 950 Pennsylvania Avenue, N.W. Washington, DC 20530 Anne Murphy, Esq. [ARGUED] United States Department of Justice Appellate Section 7644 950 Pennsylvania Avenue, N.W. Washington, DC 20530 Counsel for Appellee Fred T. Magaziner, Esq. Dechert 2929 Arch Street 18th Floor, Cira Centre Philadelphia, PA 19104 Counsel for Amicus Appellant American Civil Liberties Union of Pennsylvania Andrew G. Crocker, Esq. Electronic Frontier Foundation 815 Eddy Street San Francisco, CA 94109 Counsel for Amicus Appellant Electronic Frontier Foundation 3 ________________ OPINION ________________ SMITH, Circuit Judge. This case comes to us a second time and requires that we consider the constitutionality of the recordkeeping, labeling, and inspection requirements set forth in 18 U.S.C. §§ 2257 and 2257A (collectively, “the Statutes”) and their accompanying regulations, 28 C.F.R. §§ 75.1–75.9. Because the administrative search regime imposed by regulation violates the Fourth Amendment as applied to Plaintiffs, we will affirm in part and vacate in part the District Court’s judgment as to those claims. We will also affirm in part the District Court’s judgment that the Statutes and regulations do not violate the First Amendment. In light of our conclusion that the Government must obtain a warrant before conducting a search under the Statutes, however, we will remand to the District Court to consider whether 28 C.F.R. § 75.5(c)(1)’s additional requirement that producers make their records available for at least twenty hours per week also violates the First Amendment. I. Since 1984, Congress has criminalized both the commercial and noncommercial use of children in sexually explicit materials. See Free Speech Coal., Inc. v. Att’y Gen. (FSC I), 677 F.3d 519, 525 (3d Cir. 2012) (describing legislative efforts to criminalize child pornography). But 4 despite these direct prohibitions of using underage models in sexually explicit materials, producers of sexually explicit materials continued to utilize youthful-looking performers. See id. at 525–26 (citing Attorney General’s Commission on Pornography, Final Report, 618 (1986)). Law enforcement was ill-equipped visually to ascertain these performers’ ages, and the risk that children were still being used in pornographic materials remained. Id. Congress thus decided to place the onus on producers to collect information demonstrating that their performers were not minors. Enacted as part of the Child Protection and Obscenity Enforcement Act of 1988, Pub. L. No. 100-690, § 7513, 102 Stat. 4181, 4487, § 2257 as amended requires producers of visual depictions of “actual sexually explicit conduct” to keep “individually identifiable records” documenting the identity and age of every performer appearing in those depictions. 18 U.S.C. § 2257(a). Section 2257A, enacted as part of the Adam Walsh Child Protection and Safety Act of 2006, Pub. L. No. 109-248, § 503, 120 Stat. 587, extended similar recordkeeping requirements to producers of depictions of “simulated sexually explicit conduct.” “Sexually explicit conduct” for the purposes of both § 2257 and § 2257A consists of “(i) sexual intercourse, including genital-genital, oral-genital, anal-genital, or oral- anal, whether between persons of the same or opposite sex; (ii) bestiality; (iii) masturbation; (iv) sadistic or masochistic abuse; or (v) lascivious exhibition of the genitals or pubic area of any person.” 18 U.S.C. § 2256(2)(A); see also 28 C.F.R. § 75.1(n). Performers engage in simulated sexually explicit conduct if a “reasonable viewer” would believe that 5 the performers had engaged in the sexually explicit conduct defined above. 28 C.F.R. § 75.1(o).1 Producers of visual depictions subject to the Statutes are required to examine “an identification document” for each performer and maintain records listing each performer’s name, date of birth, and any other name that the performer has previously used. 18 U.S.C. § 2257(b); id. § 2257A(b). These records must be maintained at the producer’s “business premises,” or at any other place prescribed by regulation, and shall be made available for inspection by the Attorney General “at all reasonable times.” Id. § 2257(c); id. § 2257A(c). Producers must also “affix[] to every copy” of covered depictions “in such manner and in such form as the Attorney General shall by regulations prescribe, a statement describing where the records required . . . with respect to all performers depicted in that copy . . . may be located.” Id. § 2257(e)(1); id. § 2257A(e)(1). 1 Certain commercial producers of simulated sexually explicit depictions, along with some commercial producers of images that depict the actual lascivious exhibition of the genitals or pubic area regulated under § 2257, are exempt from these recordkeeping requirements. 18 U.S.C. § 2257A(h). These exemptions are intended to apply to industries where Congress believed that existing regulatory schemes already “adequately achieve[d] the same age-verification ends as the Statutes,” such as the mainstream motion picture and television industries. FSC I, 677 F.3d at 535 n.11; see also 152 Cong. Rec. S8012, S8027 (July 20, 2006) (statement of Sen. Patrick Leahy). 6 Detailed regulations further refine the recordkeeping and labeling requirements under the Statutes. Pursuant to these regulations, producers must maintain “a legible hard copy” of the identification documents examined for each performer, as well as a copy of each sexually explicit depiction. 28 C.F.R. § 75.2(a)(1). Producers must also generate an index tying each depiction to all names used by each performer. Id. § 75.2(a)(2)–(3); id. § 75.3. To comply with these requirements, producers are permitted to contract with a third party for record-retention purposes. Id. § 75.2(h); id. § 75.4. Regulations further specify where the statement describing the records’ location must be affixed for each copy of a sexually explicit depiction, along with the contents of that statement. Id. § 75.6; id. § 75.8. The Statutes’ general command that records be available for inspection “at all reasonable times,” 18 U.S.C. § 2257(c); id. § 2257A(c), is also governed by detailed regulations. Investigators are “authorized to enter without delay and at reasonable times any establishment of a producer where records . . . are maintained to inspect during regular working hours and at other reasonable times, and within reasonable limits and in a reasonable manner, for the purpose of determining compliance” with the Statutes. 28 C.F.R. § 75.5(a). Although inspections are to be conducted either during normal business hours or at such times that the producer “is actually conducting business” related to covered depictions, producers must nevertheless make their records available for inspection for at least twenty hours per week. Id. § 75.5(c). 7 Inspectors are further required by regulation to take several steps at the time a search is conducted to reassure producers of the lawfulness of any search. These include presenting credentials and explaining the limited nature and purpose of the inspection. Id. § 75.5(c)(2). The frequency of inspections is also circumscribed: only one inspection is permitted during any four-month period, unless law enforcement has “reasonable suspicion” that a violation has occurred. Id. § 75.5(d). Although “inspections shall be conducted so as not to unreasonably disrupt” operations, id. § 75.5(c)(3), the regulations also mandate that “[a]dvance notice of record inspections shall not be given.” Id. § 75.5(b). Failure to maintain the necessary records, to affix the necessary statement describing the records’ location to each copy of a regulated depiction, or to permit a required inspection are criminal offenses. 28 U.S.C. § 2257(f); id. § 2257A(f). First-time violators of § 2257 face a maximum sentence of five years’ incarceration, with subsequent violations punishable by imprisonment of “not more than 10 years but not less than 2 years.” Id. § 2257(i). Sentences for violations of § 2257A are capped at one year, unless the violation involves an effort to conceal a substantive offense involving the use of a minor in sexually explicit depictions, in which case the sentencing range mirrors that imposed for violations of § 2257. Id. § 2257A(i). II. Plaintiffs are a collection of individuals, commercial entities, and interest groups who are engaged in or represent others involved in the production of images covered under the 8 Statutes.2 This case first came to us following the District Court’s grant of the Government’s motion to dismiss. We held at that time that Plaintiffs stated viable as-applied and facial claims under both the First and Fourth Amendments. See FSC I, 677 F.3d at 535–46. On remand, following a bench trial on Plaintiffs’ surviving claims, the District Court concluded that the Statutes and regulations passed constitutional muster with one exception: Inspections without prior notice to examine records located in private residences 2 Specifically, these Plaintiffs are Free Speech Coalition, Inc., “a trade association representing more than 1,000 member businesses and individuals involved in the production and distribution of adult materials”; the American Society of Media Photographers, a trade association representing photographers; Thomas Hymes, “a journalist who operates a website related to the adult film industry”; Townsend Enterprises, Inc., doing business as the Sinclair Institute, “a producer and distributor of adult materials created for the purpose of educating adults about sexual health and fulfillment”; Carol Queen, “a sociologist, sexologist, and feminist sex educator”; Barbara Nitke, “a faculty member for the School of Visual Arts in New York City and a photographer”; Marie L. Levine, also known as Nina Hartley, a performer, sex educator, and producer of adult entertainment; Betty Dodson, “a sexologist, sex educator, author, and artist”; Carlin Ross, “who hosts a website with Dodson providing individuals ashamed of their genitalia with a forum for anonymously discussing and posting images of their genitalia”; and photographers Barbara Alper, David Steinberg, and Dave Levingston. FSC I, 677 F.3d at 524 n.1. 9 violated the Fourth Amendment. Free Speech Coal., Inc. v. Holder (FSC II), 957 F. Supp. 2d 564, 607–08 (E.D. Pa. 2013). But the District Court declined to issue an injunction and granted only declaratory relief. Id. at 609. In doing so, the District Court made several factual findings. The District Court found credible Plaintiffs’ testimony that “it is their sincere belief that the use of sexually explicit material is a valued artistic endeavor and also serves valued educational motives.” Id. at 583. But each Plaintiff also “consistently use[d] young-looking performers and . . . almost all of their work had a commercial or profit motive.” Id. at 584. Indeed, no Plaintiff was “an exclusive producer of sexually explicit depictions of ‘clearly mature’ adults.” Id. Nor was there any “evidence that any Plaintiff produces purely noncommercial sexual depictions or maintains records for such depictions.” Id. at 586. Further, the District Court found that the Statutes made only two of Plaintiffs’ existing or intended works “practically impossible” to create. Id. at 585. The first is Plaintiffs Carlin Ross’s and Betty Dodson’s “genital art gallery” displaying anonymous submissions of individuals’ genitalia for the purpose of demonstrating that each person’s sexual anatomy is normal. Id. at 574, 585. Submissions decreased markedly when Dodson and Ross began to require identification in response to the Statutes. The second is Plaintiff Barbara Alper’s documentary effort to photograph the adult, gay community on New York’s Fire Island engaging in anonymous sex. Id. at 573, 585. The inability of performers 10 lawfully to remain anonymous effectively defeats the purpose of that project.3 As to the general types of pornography publicly available, the District Court found that “[y]outhful-looking performers are ubiquitous in the adult entertainment industry.” Id. at 584. Testimony from Government expert Gail Dines demonstrated that “‘teen porn’4 accounts for approximately one-third of the material on pornography tube sites”—Internet portals that host a significant amount of free sexually explicit content. Id. at 586. Further, youthful- looking performers appear not only in depictions categorized as “teen porn” but also other genres nominally involving older adults. Id. at 584. The District Court found Dines’s analysis to be more methodologically rigorous than that from any of Plaintiffs’ experts. In particular, the District Court criticized Plaintiffs’ expert Daniel Linz’s reliance on “simple 3 We note that the performers would not need to show their identifications to one another. Only Alper would need to see their identifications. Thus, Alper could still photograph persons having anonymous sex, the only difference being that the performers would not be anonymous to her. Of course, if law enforcement chose to inspect Alper’s records, it would also become privy to these performers’ identities. 4 As discussed further, infra, Dines’s identification of “teen porn” involved searching specific pornographic websites for terms associated with youthful performers. In general, “teen porn” is that category of pornography that depicts individuals who could be teenagers, although commercial producers of pornography contend that the models are not actually under 18. 11 Google searches” to determine the scope of pornography publicly available. Id. at 586. The District Court also accorded “significant weight” to the testimony of Government expert Francis Biro. Id. at 586. As Biro explained, “12, 13 and 14 year olds can appear to be much older than they are because they may experience early sexual and physical maturation,” thus demonstrating “the inability to determine chronological age from visual inspections.” Id. Indeed, “[e]ven maturation experts will have a 2–5 year margin of error when trying to ascertain the age of a young adult, . . . and that margin is greater for members of the public.” Id. at 578. By contrast, the District Court viewed Plaintiffs’ experts with skepticism. Specifically, the District Court rejected testimony from Plaintiffs’ experts Michelle Drouin and Marc Zimmerman as to the prevalence of “sexting,” defined as “the sending of text messages containing sexually explicit depictions over cell phones and similar devices.” Id. at 576, 587. In particular, because Drouin’s and Zimmerman’s studies used convenience samples and respondent-driven sampling instead of random sampling, the District Court concluded that those analyses “were not worthy of weight.”5 Id. at 587. Further, because their analyses did 5 A convenience sample involves surveying “members of the population that are easily accessible to you for whatever reason.” App. 5992. Respondent-driven sampling is a form of convenience sampling and involves incentivizing volunteers to recruit others to participate in the study. App. 5998. 12 not quantify the amount of sexting involving “images of intercourse, masturbation, or a ‘lascivious display’ of genitals” that would be covered under the Statutes, Drouin’s and Zimmerman’s testimony failed to demonstrate any overbreadth related to sexting. Id. Finally, the District Court also made findings regarding the searches conducted to ensure compliance with the Statutes. Although 29 inspections had been conducted since 2006, “the inspections program at the FBI has been completely shut down” since early 2008 and no plans are in place to recommence searches. Id. at 579–88. But those inspections that were conducted “involved physical intrusions” into nonpublic spaces of both businesses and residences, including internal offices, supply rooms, kitchens, conference rooms, dining rooms, and garages. Id. at 588. And despite the regulations’ express prohibition, “[a]dvance notice was given on several occasions,” without evidence that the integrity of any search was undermined. Id. at 588–89. Indeed, even the testifying FBI agents “believed it would be very difficult if not impossible to fabricate the records required by the Statutes in a 24-hour period.” Id. at 589. III. The District Court had jurisdiction pursuant to 28 U.S.C. § 1331. We have jurisdiction pursuant to 28 U.S.C. § 1291. We review legal questions de novo, including the constitutionality of the federal statutes and regulations at issue here. ACLU v. Mukasey, 534 F.3d 181, 186 (3d Cir. 2008). The court’s factual findings following a bench trial are typically reviewed for clear error. Post v. St. Paul 13 Travelers Ins. Co., 691 F.3d 500, 514 (3d Cir. 2012) (citing Am. Soc’y for Testing & Materials v. Corrpro Cos., 478 F.3d 557, 566 (3d Cir. 2007)). But “[i]n the First Amendment context, reviewing courts have a duty to engage in a searching, independent factual review of the full record” to the extent any factual findings are relevant to the First Amendment constitutional standard. ACLU, 534 F.3d at 186 (alteration in original) (quoting United States v. Scarfo, 263 F.3d 80, 91 (3d Cir. 2001)). This independent review “is not equivalent to a ‘de novo’ review of the ultimate judgment itself” but is necessary to ensure “that the judgment does not constitute a forbidden intrusion on the field of free expression.” Bose Corp. v. Consumers Union of U.S., Inc., 466 U.S. 485, 508, 514 n.31 (1984) (quoting N.Y. Times Co. v. Sullivan, 376 U.S. 254, 285 (1964)). IV. A. We first consider Plaintiffs’ as-applied challenge under the First Amendment. In that regard, we have already determined that the Statutes are content neutral. FSC I, 677 F.3d at 533–35. Accordingly, we apply intermediate scrutiny to evaluate the burdens imposed on Plaintiffs’ free-speech rights. Id. at 535. “A statute satisfies intermediate scrutiny where it: (1) advances a ‘substantial’ governmental interest; (2) does not ‘burden substantially more speech than is necessary’ (i.e., the statute must be narrowly tailored); and (3) leaves open ‘ample alternative channels for communication.’” Id. (quoting Ward v. Rock Against Racism, 491 U.S. 781, 791, 798–800 (1989)). 14 In FSC I, we agreed with the District Court that two of these prongs were satisfied as a matter of law. Specifically, we held that “the Statutes clearly advance a substantial governmental interest—protecting children from sexual exploitation by pornographers.” Id. They do so in at least four specific ways: (1) they ensure that primary producers of sexually explicit expression confirm the ages of their performers prior to filming; (2) they permit secondary producers that publish the depictions to verify that the performers were not children; (3) they prevent children from passing themselves off as adults; and (4) they aid law enforcement and eliminate subjective disputes with producers over whether the producer should have verified the age of a particular performer. Id. Further, “the Statutes leave open ample alternative channels for communication” given that “[t]he Statutes regulate recordkeeping and labeling procedures and do not ban or otherwise limit speech.” Id. at 536 n.13. In this appeal, we are left to determine whether the District Court properly concluded that the Government has met its burden to prove that the Statutes and their implementing regulations are narrowly tailored. See Startzell v. City of Phila., 533 F.3d 183, 201 (3d Cir. 2008) (for as-applied First Amendment challenges, “[t]he burden is on the [government] to demonstrate the constitutionality of its actions”). Narrow tailoring does not require that the regulation be “the least restrictive or least intrusive” means of achieving 15 “the government’s legitimate, content-neutral interests.” Ward, 491 U.S. at 798. Instead, “[n]arrow tailoring is satisfied where the statute at issue does not ‘burden substantially more speech than is necessary to further the government’s legitimate interests.’” FSC I, 677 F.3d at 536 (quoting Ward, 491 U.S. at 799). In this case, part of our inquiry thus involves comparing “the amount of Plaintiffs’ constitutionally-protected speech that does not implicate the government’s interest in protecting children” with “the amount of Plaintiffs’ speech that implicates the government’s interest.” Id. at 537. But we must be careful not to reduce our narrow- tailoring inquiry under intermediate scrutiny to a rigid mathematical formula or purely quantitative assessment. See Henderson v. Lujan, 964 F.2d 1179, 1184 (D.C. Cir. 1992) (“Despite the seemingly mathematical character of the metaphor, the Supreme Court in fact applies [narrow tailoring] as a balancing test . . . .”). Our analysis when applying intermediate scrutiny “always encompasses some balancing of the state interest and the means used to effectuate that interest,” and “varies to some extent from context to context, and case to case.” Bartnicki v. Vopper, 200 F.3d 109, 124 (3d Cir. 1999), aff’d, 532 U.S. 514 (2001). In keeping with this view of the narrow-tailoring inquiry as a balancing test, we ask whether “a substantial portion of the burden on speech does not serve to advance [the government’s] goals.” McCullen v. Coakley, 134 S. Ct. 2518, 2535 (2014) (quoting Ward, 491 U.S. at 799). Indeed, “unless we know the extent to which the [regulations] in fact interfere with protected speech, we cannot say whether they 16 suppress ‘substantially more speech than . . . necessary.’” Turner Broad. Sys., Inc. v. FCC, 512 U.S. 622, 668 (1994) (second alteration in original) (quoting Ward, 491 U.S. at 799); see also Bartnicki, 200 F.3d at 129 (considering “the practical impact” of the regulation). Thus, assessing the scope of the burden on speech looks not just to the quantity of speech burdened, but also to the qualitative nature of that burden. And even if a significant amount of speech is burdened in a manner that is not strictly necessary to fulfill the Government’s stated interest, a regulation may in some circumstances still be sufficiently narrowly tailored if the nature of the actual burden imposed on that speech is minimal. Our consideration here of the nature of the burden imposed in addition to the quantity of speech that a regulation burdens is not novel. In Ward v. Rock Against Racism, the Supreme Court considered whether a regulation requiring performers in a public bandshell to use city-provided sound equipment and technicians survived intermediate scrutiny. 491 U.S. at 784. Holding that the regulation was sufficiently narrowly tailored, the Court rejected the plaintiff’s contention that such a regulation “sweeps far more broadly than is necessary to further the city’s legitimate concern with sound volume.” Id. at 801. In doing so, the Court found the lack of a “substantial deleterious effect on the ability of bandshell performers to achieve the quality of sound they desired” to favor the conclusion that the regulation was narrowly tailored. Id. at 801–02. Because the regulation did not function as a ban on speech, but instead “focuse[d] on the source of the evils the city s[ought] to eliminate . . . and eliminate[d] them 17 without at the same time banning or significantly restricting a substantial quantity of speech that does not create the same evils,” the regulation exemplified “the essence of narrow tailoring.” Id. at 799 n.7. Further, the nature of the Government’s interest and the extent to which that interest is advanced also factors into the narrow-tailoring calculus. Whether a content-neutral regulation is narrowly tailored “does not turn on a judge’s agreement with the responsible decisionmaker concerning the most appropriate method for promoting significant government interests or the degree to which those interests should be promoted.” Id. at 799 (quoting United States v. Albertini, 472 U.S. 675, 689 (1985)) (internal quotation marks omitted). Indeed, the Government has “some discretion in deciding how best to achieve its legitimate purposes.” Brown v. City of Pittsburgh, 586 F.3d 263, 277 (3d Cir. 2009). Nevertheless, “[a] restriction cannot be ‘narrowly tailored’ in the abstract; it must be tailored to the particular government interest asserted.” McTernan v. City of York, Pa., 564 F.3d 636, 656 (3d Cir. 2009). Thus, “[t]o meet the requirement of narrow tailoring, the government must demonstrate that alternative measures that burden substantially less speech would fail to achieve the government’s interests, not simply that the chosen route is easier.” McCullen, 134 S. Ct. at 2540. B. These principles not only guide our analysis of the merits of Plaintiffs’ as-applied First Amendment claim, but also impact the Government’s threshold challenge as to Free 18 Speech Coalition’s (FSC) and the American Society of Media Photographers’ (ASMP) associational standing to bring an as- applied claim on behalf of the entire adult film industry. Associational standing requires that at least one of the organizational plaintiffs “demonstrate that ‘(a) its members would otherwise have standing to sue in their own right; (b) the interests it seeks to protect are germane to the organization’s purpose; and (c) neither the claim asserted nor the relief requested requires the participation of individual members in the lawsuit.’” Pa. Psychiatric Soc’y v. Green Spring Health Servs., Inc., 280 F.3d 278, 283 (3d Cir. 2002) (quoting Hunt v. Wash. State Apple Adver. Comm’n, 432 U.S. 333, 343 (1977)). But neither FSC nor ASMP represents “the adult film industry” as a whole. Instead, their members comprise various segments of that industry. And those individual members’ participation is necessary to assess properly FSC’s and ASMP’s as-applied First Amendment claims. Specifically, we must examine whether the Statutes and regulations are sufficiently circumscribed as they apply to the specific conduct of FSC’s and ASMP’s members. See Bd. of Trs. of State Univ. of N.Y. v. Fox, 492 U.S. 469, 482 (1989) (a plaintiff alleging a statute is not narrowly tailored “asserts that the acts of his that are the subject of the litigation fall outside what a properly drawn prohibition could cover”). And as our description of the law governing narrow tailoring makes clear, whether the Statutes and regulations survive intermediate scrutiny as applied to each producer of sexually explicit images turns on the degree to which that individual producer’s speech is unnecessarily burdened. Indeed, the 19 Statutes might be narrowly tailored as to some of FSC’s and ASMP’s members but not others depending upon the nature of each member’s speech. Identifying those members for whom the Statutes may be unconstitutional thus requires an individualized inquiry that fails to satisfy the requirements for associational standing. Cf. Harris v. McRae, 448 U.S. 297, 321 (1980) (individualized nature of a Free Exercise claim negated organizational standing); Bano v. Union Carbide Corp., 361 F.3d 696, 714 (2d Cir. 2004) (no organizational standing where “‘the fact and extent’ of the injury that gives rise to the claims for injunctive relief ‘would require individualized proof’” (quoting Warth v. Seldin, 422 U.S. 490, 515–16 (1975))).6 6 We note that requests for prospective injunctive relief like that sought here typically “do not require participation by individual association members,” Hosp. Council of W. Pa. v. City of Pittsburgh, 949 F.2d 83, 89 (3d Cir. 1991), because such a blanket remedy “inure[s] to the benefit of those members of the association actually injured,” Warth, 422 U.S. at 515. But even if the remedy does not require individual proof, “[c]ourts ‘also must examine the claims asserted to determine whether they require individual participation.’” Alliance for Open Soc’y Int’l., Inc. v. U.S. Agency for Int’l Dev., 651 F.3d 218, 229 (2d Cir. 2011) (quoting Rent Stabilization Ass’n v. Dinkins, 5 F.3d 591, 596 (2d Cir. 1993)), aff’d, 133 S. Ct. 2321 (2013). Here, the need for extensive individual participation to assess how each member’s speech is impacted precludes associational standing despite the nature of the relief sought. 20 To be sure, FSC’s members comprise individuals and businesses across many facets of the adult film industry, including those involved in the creation, distribution, and sale of both live and prerecorded sexually explicit materials. Some of those members, like Vivid Video, Wicked Pictures, K-Beech Video, and Darkside Productions, are large-scale producers of commercial pornography. And ASMP represents some 400 photographers whose work involves sexually explicit images. But even if FSC’s and ASMP’s members collectively produce a significant portion of the works generated by the adult film industry, aggregating that industry’s speech in toto is an improper method for identifying the burdens imposed on specific members. Generalized statements regarding the adult film industry’s speech cannot replace the individualized inquiry required, and FSC and ASMP lack associational standing to bring an as- applied First Amendment claim on behalf of their members. C. Our analysis of Plaintiffs’ as-applied First Amendment challenge is therefore limited to those Plaintiffs engaged in the actual production of images covered under the Statutes. We examine first the nature of the burden placed on these Plaintiffs. That the Statutes’ requirements here do not operate as a ban on the speech in which Plaintiffs would engage weighs in favor of finding those regulations narrowly tailored. As the District Court found, only a very limited number of Plaintiffs’ works are seriously impacted. Plaintiffs are still free in large measure to produce the very same speech that they would produce absent the Statutes’ recordkeeping requirements. See Hill v. Colorado, 530 U.S. 703, 726 (2000) 21 (“[W]hen a content-neutral regulation does not entirely foreclose any means of communication, it may satisfy the tailoring requirement even though it is not the least restrictive or least intrusive means of serving the statutory goal.”). We acknowledge that compliance with the Statutes and regulations, as with any regulatory requirement, is more costly than if those requirements did not exist. Plaintiffs must create and maintain, or contract with a third party to maintain, records pertaining to each performer depicted in a sexually explicit image. And they must make those records available to law enforcement upon request. But these requirements are 22 not so onerous as to become prohibitive.7 Indeed, as the District Court observed, even plaintiff Sinclair Institute, “the world’s largest producer[] and distributor[] of adult sexual 7 We recognize that some Plaintiffs testified to the belief that every unpublished image created during a photoshoot—which could amount to hundreds of prints—needed to be labeled individually to comply with the Statutes. The District Court dismissed these concerns because “the regulations do not appear to require a label be affixed until a photograph is made publicly available.” FSC II, 957 F. Supp. 2d at 591 n.15 (citing 28 C.F.R. § 75.8(a)–(f)). Whether that distinction is tenable, we note that for many works, the regulations do not require every image to be labeled. For example, books, magazines, and periodicals need be labeled only “on the first page that appears after the front cover or on the page on which copyright information appears.” 28 C.F.R. § 75.8(a). And “a digital video disc (DVD) containing multiple depictions is a single matter for which the statement may be located in a single place covering all depictions on the DVD.” Id. § 75.8(e). To the extent unpublished images do not fall under any of the express categories in the regulations, “the statement is to be prominently displayed consistent with the manner of display required for the aforementioned categories.” Id. § 75.8(f). Therefore, it seems the regulations would require only a single label for a compilation of unpublished images from a photoshoot, consistent with the requirement for books, magazines, and DVDs. We need not resolve this dispute, however, because even if such labels must be affixed to every print in a photoshoot, we would still find the Statutes and regulations to be narrowly tailored. 23 education and health media,” spends only $75,000 per year to comply with the Statutes despite generating millions in revenue from the production of sexually explicit images. See FSC II, 957 F. Supp. 2d at 575 (alterations in original). And several of the individual plaintiffs, including David Steinberg who maintains the required records in his home office and Marie Levine who maintains records through a third-party custodian, have not found compliance with the Statutes to be financially prohibitive. That the burden imposed on Plaintiffs here is relatively minimal supports the conclusion that the Statutes are narrowly tailored. See Ward, 491 U.S. at 799 n.7. Some Plaintiffs testified that they modified their works in response to the Statutes’ requirements, while others produce fewer or different images than they would have if the Statutes did not exist. For example, journalist Thomas Hymes refuses to upload certain images to his website to avoid running afoul of the Statutes. Photographer Dave Levingston avoids publishing certain images that he deems could be interpreted as depicting a simulated sexual act and no longer photographs models in such poses. And the burden of collecting identification records for all performers has discouraged additional Plaintiffs from producing still other works, along with discouraging some potential adult performers from appearing in sexual depictions. But the fact that some Plaintiffs are avoiding publishing certain images is not directly attributable to the Statutes and regulations themselves and is not equivalent to a governmental ban. We are cognizant that “courts must be wary that taxes, regulatory laws, and other laws that impose financial burdens are not used to undermine . . . freedom of speech.” Pitt News v. 24 Pappert, 379 F.3d 96, 110 (3d Cir. 2004). But in light of several Plaintiffs’ compliance with the Statutes without incident or prohibitive financial burdens, Levingston’s and Hymes’s choices to forego publication of certain images amount to nothing more than personal decisions to avoid costs that they could otherwise bear. Cf. id. (“A business in the communications field cannot escape its obligation to comply with generally applicable laws on the ground that the cost of compliance would be prohibitive.”). On balance, this evidence does not demonstrate that the Statutes’ objective burdens for the purpose of a narrow-tailoring analysis are any greater than we have already discussed. D. Having considered the nature of the burden imposed on Plaintiffs, we turn next to the amount of each Plaintiffs’ speech needlessly impacted. Here, the Government’s interest in enforcing the Statutes is to prevent producers of sexually explicit materials from depicting minor performers, either purposefully or inadvertently. Achieving this interest becomes particularly complicated given the difficulty of discerning a potential performer’s age by sight alone. But as we stated when this case first came to us, burdening “speech involving performers who are obviously adults” does not advance the Government’s interest in protecting children. FSC I, 677 F.3d at 537. Requiring identification and recordkeeping for clearly mature performers does nothing to prevent children from appearing in sexually explicit materials because, by definition, a minor could not be mistaken for a clearly mature adult. 25 The Government takes the position that the evidence produced at trial demonstrates that any attempt to identify a class of clearly mature adults exempt from the Statutes’ reach would undermine the Statutes’ effectiveness. In the Government’s view, reliably identifying a performer as clearly mature by sight is a fruitless endeavor given the different rates at which individuals visibly mature along with minors’ ability to appear older through makeup and other artificial means. Thus, the argument goes, anything less than a uniform identification requirement would leave children at risk. In support of this argument, the Government’s pubertal maturation expert, Francis Biro, testified that “the average person, without training and experience in maturation assessment” would “have difficulty in determining someone’s age by visual inspection.” App. 5467–68. But Biro also testified that it is “generally true[,] [b]ut not always true” that adults who are 25 years of age or older will not be mistaken for minors under age 18. App. 5492– 5493. And “the vast majority” of adults 30 years of age or older could not be mistaken for a minor. App. 5493. Further, Biro conceded that one could not confuse a 60-year-old for a minor. App. 5491. These observations support our previously stated view that the Statutes’ application to producers of only “an illustrated sex manual for the elderly”—but who did not employ youthful-looking performers—may be unconstitutional. FSC I, 677 F.3d at 537 (quoting Am. Library Ass’n v. Reno, 33 F.3d 78, 90 (D.C. Cir. 1994)). Thus, we reject the Government’s contention that age verification for all performers regardless of their actual age 26 always furthers the Government’s interest in preventing the sexual exploitation of minors. While a universal recordkeeping requirement would be easier for the Government to enforce, ease of enforcement is not the touchstone for narrow tailoring. In McCullen v. Coakley, the Supreme Court considered whether a law proscribing “knowingly stand[ing] on a ‘public way or sidewalk’ within 35 feet of an entrance or driveway to any place, other than a hospital, where abortions are performed” survived intermediate scrutiny. 134 S. Ct. at 2525 (quoting Mass. Gen. Laws ch. 266 § 120E1/2(a), (b)). In holding that this law was not narrowly tailored, the Court observed that “[a] painted line on the sidewalk is easy to enforce, but the prime objective of the First Amendment is not efficiency.” Id. at 2540. Instead, “[t]o meet the requirement of narrow tailoring, the government must demonstrate that alternative measures that burden substantially less speech would fail to achieve the government’s interests, not simply that the chosen route is easier.” Id. (emphasis added). Here, given that the Government’s own expert testified that at a certain advanced age, no individual could be mistaken for a minor, the Government has not established that imposing some age cutoff would necessarily undermine the Statutes’ effectiveness in preventing the exploitation of children. Preventing all erroneous age determinations does not advance the Government’s interest in combatting child pornography where even an error would not run the risk that a minor would appear in sexually explicit materials. For instance, where an individual appears 60 years old, no one could mistake that same individual for a minor under 18. 27 Mistakes could be made about that person’s specific age, but not his/her minor status. Requiring age-verification records for an apparent 60-year old is therefore a burden that does nothing to protect children. Thus, the difficulties in accurately calculating age by sight alone justifies some of the Statutes’ prophylactic reach, but the Statutes’ burdens do not advance the Government’s interest when imposed on performers whom no reasonable person could mistake for a minor. This observation does not mean, however, that the Statutes are not narrowly tailored as applied to these Plaintiffs. Indeed, time and again we have stated that under intermediate scrutiny, the Government need not employ the least restrictive or least intrusive means. See, e.g., King v. Governor of N.J., 767 F.3d 216, 239 (3d Cir. 2014); FSC I, 677 F.3d at 535; McTernan, 564 F.3d at 655. “So long as the means chosen are not substantially broader than necessary to achieve the government’s interest, . . . the regulation will not be invalid simply because a court concludes that the government’s interest could be adequately served by some less-speech-restrictive alternative.” Ward, 491 U.S. at 800 (emphasis added). And in this case, “[t]he Government must be allowed to paint with a reasonably broad brush if it is to cover depictions of all performers who might conceivably have been minors at the time they were photographed or videotaped.” Am. Library Ass’n, 33 F.3d at 90. But neither side successfully established at trial where the line between “clearly mature” and “possibly underage” can effectively be drawn. Plaintiffs, relying on Biro’s 28 testimony,8 observe that most minors would not be mistaken for someone over the age of 25. Using that cutoff, Plaintiffs urge that requiring identification for anyone who appears over 25 imposes burdens that do not advance the Government’s interests. But Biro’s statement that generally most minors could not be mistaken for a 25-year-old adult does not establish that the Government’s interests are not furthered by requiring identification for performers over age 25. To the contrary, Biro’s testimony establishes that the rare minor could appear up to 30 years old.9 Failing to require producers to check identification for such individuals would therefore render the Statutes less effective in preventing child pornography. Thus, at the very least, comparing the use of performers above and below age 25 as Plaintiffs urge does 8 Plaintiffs did not offer their own expert on pubertal maturation, and do not otherwise challenge Biro’s testimony regarding visual determinations of age. Further, the District Court found Biro’s testimony on these topics to be credible. FSC II, 957 F. Supp. at 586. After conducting an independent review of that testimony, we see no reason to reject Biro’s findings. 9 We express no view as to whether minors in fact could appear to be over 30 years old, particularly given the possibility of makeup and other artificial enhancements. Indeed, Biro himself equivocated on this point. However, we need not address whether the Government in a different case and on a different record can prove that requiring identification even for performers who appear over 30 helps protect children. 29 not advance their argument that the Statutes are not narrowly tailored. At all events, we need not identify the precise age at which no minor could reasonably be mistaken for an adult of that age to conduct a narrowing-tailoring analysis. Given that it is the Government’s burden to prove that the Statutes are narrowly tailored, we can instead look to which burdens the Government has established in fact advance the goal of preventing the sexual exploitation of minors. As noted above, Biro’s testimony supports a finding that requiring identification for performers up to age 30 implicates the Government’s interest in protecting children. Yet even if we accept that the Government proved no more than that and that requiring recordkeeping for individuals 30 and older does not advance the Government’s interest, the evidence indeed demonstrates that a significant proportion of Plaintiffs’ work falls squarely within the Statutes’ permissible scope. Specifically, of the models employed by Dodson and Ross, 45% of them are under 30 years of age. Levine’s under-30 models account for 40.3%. Similarly, 47.37% of Nitke’s models are under 30. Levingston’s number 60%. And 33.97% of the Sinclair Institute’s models were under 30. Finally, based on self-reported figures, 24% of Steinberg’s models were under 30 while the “vast majority” of participants in Carol Queen’s live-streamed masturbation 30 fundraiser were in their 30s and 40s along with “10 to 12 percent” under age 25. App. 5973.10 Undoubtedly, these figures demonstrate that the number of performers to whom the Statutes apply, yet for whom requiring identification does not protect children, is not insignificant. But equally clear is that each Plaintiff’s work depicts a substantial number of individuals for whom requiring identification does promote the Government’s interests. And as noted previously, the qualitative burdens imposed on Plaintiffs are not so great that any works beyond one project of Dodson’s and Ross’s and one of Alper’s are 10 The record does not reflect any age breakdowns for models utilized by Barbara Alper, but the burdens she identified centered upon her inability to permit her subjects to remain anonymous. Alper assures us that each of her subjects would be an adult, but not that they would be clearly mature. Nor does the record otherwise reflect what proportion of her photographic efforts would have been limited to clearly mature adults. Therefore, we are unable to identify which burdens advance the Government’s interests with respect to her desired speech and which do not. Suffice it to say, however, that the Government’s interest in ensuring that minors are not sexually exploited is advanced when completely anonymous sexual participants whom we have no reason to believe are over 18 must verify their ages before appearing in sexually explicit materials. 31 seriously impacted.11 Thus, although some Plaintiffs use smaller numbers of under-30 performers, the minimal qualitative burdens of storing identification records imposed on even these Plaintiffs still supports a finding that the Statutes are narrowly tailored. Another crucial point here is that Plaintiffs do not face a substantial additional burden attributable to keeping records for clearly mature performers on top of the records they must maintain for young performers. Instead, most of the burden Plaintiffs face under the Statutes is due to the procedures they must put in place to store, organize, and make available records for performers generally. These startup costs associated with creating a recordkeeping system under the Statutes, including the costs of creating indexes, advance the Government’s interests in preventing the sexual exploitation of children. Collecting additional identification for the clearly mature performers that each Plaintiff also employs and inserting them into this system does not impose a significant additional burden. For example, Plaintiffs are not required to create a separate electronic database for clearly mature adults. Instead, any clearly mature performers would be just one more data point in a preexisting recordkeeping 11 Moreover, projects like these illustrate precisely why Congress enacted the Statutes. Regarding Dodson’s and Ross’s project, it is especially difficult to discern whether a person is a minor based on an isolated picture of that person’s genitals. And as noted above, we have no reason to conclude that Alper’s subjects would clearly be mature. The difficulty in distinguishing minors from adults that justifies the Statutes’ reach thus pervades both of these projects. 32 scheme. Thus, the additional burdens arising from collecting identification from clearly mature adults does not establish that a “substantial portion of the burden on speech does not serve to advance [the Government’s] goals.” McCullen, 134 S. Ct. at 2535 (quoting Ward, 491 U.S. at 799). To the contrary, most of the burden Plaintiffs incur through compliance with the Statutes is implicated by the Government’s interest in protecting children. Accordingly, because Plaintiffs each employ a substantial number of youthful-looking models, the qualitative burden to comply with the Statutes is minimal and prohibits none of their speech, and because most of the burden Plaintiffs face in establishing an identification and recordkeeping system accessible by law enforcement advances the Government’s interest in combatting child pornography, we hold that the Statutes and regulations, with 33 one possible exception,12 are narrowly tailored as applied to Plaintiffs. 12 As discussed infra, we also hold that 28 C.F.R. § 75.5 as applied to Plaintiffs violates the Fourth Amendment insofar as it authorizes warrantless searches. This holding raises doubt about the constitutionality under the First Amendment of § 75.5’s command that producers also make their records available for at least twenty hours per week during pre- established periods. See id. § 75.5(c)(1). Specifically, we question whether this requirement is narrowly tailored as applied to Plaintiffs given that a warrant will allow the police to obtain the records regardless of a producer’s or third-party custodian’s specific availability to permit execution of the warrant. To require Plaintiffs, many of whom are niche participants in the adult entertainment industry, to incur the additional burden of contracting with a third party or remaining on standby themselves for twenty hours per week so that the Government may execute a warrant at preordained times seems to do little to advance the Government’s interest in protecting children. 34 As a final point on this topic, we note that we do not have in this case any producer of sexually explicit images that exclusively employs clearly mature adults, however defined. No Plaintiff therefore meets the prototypical example we identified of an individual who produces an “illustrated sex manual for the elderly.” Nor do any of the plaintiffs in this litigation produce only images intended for private use rather than public distribution. See Connection Distrib. Co. v. Holder, 557 F.3d 321, 337–38 (6th Cir. 2009) (en banc) (assuming but not deciding that application of the law to “a couple who produced, but never distributed, a home video or photograph of themselves engaging in sexually explicit conduct” would be unconstitutional). Whether the Statutes and regulations may be constitutionally applied to individuals Nor did the Government argue that the twenty-hour requirement itself helped protect children. Instead, the Government conceded that the records at issue here could not be fabricated on short notice. And the Government’s experience that conducting searches after providing notice did not undermine its inspection efforts under the Statutes further bolsters the conclusion that any legitimate purpose that the twenty-hour requirement might serve is attenuated. But because the parties focused upon the amount of Plaintiffs’ speech that is burdened rather than on the specific burden that the twenty-hour requirement imposes relative to the Government’s legitimate interests, we think the prudent course is for us to remand to the District Court for further consideration of § 75.5(c)(1)’s First Amendment implications in light of our resolution of Plaintiffs’ Fourth Amendment claim. 35 falling in either of those categories are therefore questions we need not reach. V. We turn next to Plaintiffs’ claim under the First Amendment that the Statutes are facially overbroad. This facial challenge requires consideration of four factors: “(1) ‘the number of valid applications’ of the statute; (2) ‘the historic or likely frequency of conceivably impermissible applications’; (3) ‘the nature of the activity or conduct sought to be regulated’; and (4) ‘the nature of the state interest underlying the regulation.’” FSC I, 677 F.3d at 537–38 (quoting Gibson v. Mayor & Council of Wilmington, 355 F.3d 215, 226 (3d Cir. 2004)). Demonstrating a single impermissible application will not invalidate a law on its face. “Instead, a law may be invalidated as overbroad only if ‘a substantial number of its applications are unconstitutional, judged in relation to the statute’s plainly legitimate sweep.’” Id. at 537 (quoting Wash. State Grange v. Wash. State Republican Party, 552 U.S. 442, 449 n.6 (2008)). Unlike an as-applied challenge, the burden falls upon Plaintiffs to demonstrate the Statutes’ facial overbreadth. Virginia v. Hicks, 539 U.S. 113, 122 (2003). Evidence demonstrating that the conceivably impermissible applications of a statute are in reality widespread is probative of overbreadth. Such evidence assists the court in not “go[ing] beyond the statute’s facial requirements and speculat[ing] about ‘hypothetical’ or ‘imaginary’ cases.” Wash. State Grange, 552 U.S. at 449–50. Indeed, “particularly where conduct and not merely speech is 36 involved, . . . the overbreadth of a statute must not only be real, but substantial as well, judged in relation to the statute’s plainly legitimate sweep.” Hill, 530 U.S. at 732 (quoting Broadrick v. Oklahoma, 413 U.S. 601, 615 (1973)) (internal quotation marks omitted); see also Members of City Council of L.A. v. Taxpayers for Vincent, 466 U.S. 789, 800 (1984) (“[T]he mere fact that one can conceive of some impermissible applications of a statute is not sufficient to render it susceptible to an overbreadth challenge.”); Gibson, 355 F.3d at 226 (discounting certain hypothetical scenarios as “more than slightly unrealistic”). But although we must consider the relative quantities of valid and invalid applications of the Statutes, here too we must be careful not to reduce the facial overbreadth analysis to a mathematical calculation or numerical comparison. See Aiello v. City of Wilmington, Del., 623 F.2d 845, 854 (3d Cir. 1980) (“The question is a matter of degree; it will never be possible to say that a ratio of one invalid to nine valid applications makes a law substantially overbroad . . . .” (alteration in original) (quoting Magill v. Lynch, 560 F.2d 22, 30 (1st Cir. 1977))). Instead, determining whether a statute is facially unconstitutional requires “as much in the way of judgment” as it does “a comparison between the constitutional and unconstitutional applications of a law.” Connection Distrib. Co., 557 F.3d at 340. Thus, “[s]ubstantial overbreadth involves not just an inquiry into the legitimate and illegitimate sweep of a statute; it also involves an inquiry into the ‘absolute’ nature of a law’s suppression of speech.” Id. It is not as simple as “placing . . . the number of overall applications of the statute in the denominator and the 37 number of unconstitutional applications of the statute in the numerator.” Id. A. We first compare “the amount of speech that implicates the government’s interest in protecting children” with “the amount of speech that is burdened but does not further the government’s interest” in an effort to “weigh the legitimate versus problematic applications of the Statutes.” FSC I, 677 F.3d at 538. To do so requires identifying the Statutes’ full scope. United States v. Williams, 553 U.S. 285, 293 (2008) (“[I]t is impossible to determine whether a statute reaches too far without first knowing what the statute covers.”). We previously rejected the Government’s proposed limiting construction confining the Statutes’ application to commercial images “intended for sale or trade.” FSC I, 677 F.3d at 538. Instead, the Statutes’ plain text “makes clear that they apply broadly to all producers of actual or simulated sexually explicit depictions regardless of whether those depictions were created for the purpose of sale or trade.” Id. at 539. Accordingly, the Statutes reach essentially the entire universe of sexually explicit images, “including private, noncommercial depictions created and viewed by adults in their homes.” Id. at 538. Yet although the Statutes’ reach is expansive, the valid applications of the Statutes are also plentiful. According to Gail Dines, the Government’s expert on this topic, the top three pornographic Internet websites contain 17.97 million 38 pages containing words13 clearly related to young adults. This figure amounts to 34.2% of all pages within these pornographic sites. We do not suggest that Dines’s use of Internet searches of specific websites is a precise barometer of the universe of pornography depicting youthful-looking adults publicly available. Indeed, they do not account for the amount of sexually explicit materials featuring young adults available in print or other offline media. But by limiting her searches to websites hosting pornographic material, her results are a reflection of the massive amount of pornographic material depicting youthful-looking performers available on those websites. And those results support the proposition that the legitimate sweep of the Statutes is vast. Further, identifying the amount of pornography featuring youthful-looking performers by focusing only upon the labels attached to pornographic images is underinclusive of the amount of pornography for which requiring identification and recordkeeping advances the Government’s interests. Youthful-looking performers appear across pornographic genres regardless of whether the material is expressly categorized as featuring young adults. Even materials labeled as featuring mature adults often feature a young performer engaging in sexual intercourse with an older person. And after examining all 61 categories of pornographic material on a top pornographic website—at the time, the 40th most visited website in the United States— Dines found that the “overriding image is of a youthful looking woman.” App. 5517. So even Dines’s 17.97 million- 13 Dines’s search terms were “teen,” “young,” “college,” and “daughter.” App. 5016. 39 page estimate does not represent the full swath of sexually explicit materials to which the Statutes validly apply. Given the sizeable quantity of the Statutes’ valid applications, Plaintiffs face a difficult task to show that the Statutes are substantially overbroad. Regarding depictions of clearly mature adults, Plaintiffs’ evidence of their own works, along with examples from other producers in the record, supports the proposition that there exists at least some non- negligible quantum of these kinds of images. But Plaintiffs fail in their attempts to demonstrate that the universe of sexually explicit images depicting only clearly mature adults outnumbers images depicting youthful-looking performers. To that end, Daniel Linz, one of Plaintiffs’ experts, conducted several Internet searches using Google. He found just 28 million hits for “teen pron [sic]” using simple Google searches, compared with 1.36 billion hits for “porn” alone. App. 5894. Linz testified that he used the term “teen pron” instead of “teen porn” because, in his words, “Google is very sensitive about the possibility of typing in the term child porn, so this is used as a surrogate to allow for the widest possible search.” App. 5884. When searching for “teen porn” properly spelled, his results came out to 136 million hits. But Linz offers no reason for us to accept these searches as an accurate representation of the amount of pornography available on the Internet that depicts apparent teenagers. This is especially problematic for Plaintiffs given that it is their burden, not the Government’s, to prove the Statutes’ overbreadth. Indeed, we are skeptical that Linz’s comparisons between Google results for terms like “teen porn” and results for “porn” in any way reflect the amount of 40 pornography in those genres given that neither of these searches account for the fact that many of these search results will not contain pornographic images at all. For example, news items or other written work regarding pornography from websites that do not host pornographic materials would be included in Linz’s results. Further, no party presented data on the pornography available on peer-to-peer networks or other networks that Google does not catalog. Nor does the record reflect how Google might filter search results containing pornography depicting youthful adults, a concern that motivated Linz to search the term “teen pron” instead of “teen porn” in the first place. These flaws significantly undermine the reliability of Linz’s analysis. Further, the Government presents evidence that their own Internet-wide Google searches for “teen porn” (accurately spelled) and “porn” reflect 235 million and 818 million hits, respectively. App. 5018–20. This is an enormous discrepancy with Linz’s results. And as Dines pointed out, youthful adults appear in all categories of pornography, not just “teen porn,” making the attempt to estimate the amount of sexually explicit depictions of youthful adults using categorical search terms particularly foolhardy. For these reasons, we agree with the District Court that Linz’s web searches do nothing to prove the Statutes’ potential overbreadth. We next consider Plaintiffs’ evidence of the prevalence of sexually explicit images created for private use, 41 as opposed to for sale or trade.14 Such images, because of their private nature, are difficult to quantify. Nevertheless, one of Plaintiffs’ experts on these topics, Michelle Drouin,15 established that there may be a significant number of private sexually explicit images shared between young adults. Drouin’s work, based on self-selected samples of university students, found that 54% of students in committed relationships had sent “sexually explicit pictures or videos” to their partners. App. 973, 5600. What qualified as “sexually explicit” was left undefined, and it is not clear how many of 14 The Government does not argue that burdening producers of such images advances its legitimate interests in preventing child pornography. But we note that simply because an image was originally intended to be private does not necessarily mean that it will remain so, or that one of the participants in the sexually explicit image might not have a change of heart and make the image available for sale or trade. However, whether these possibilities mean that the Government has a legitimate interest in requiring recordkeeping and labeling for even private images is not an issue we need address. 15 Marc Zimmermann also testified on behalf of Plaintiffs, but his work was less probative. His study asked whether participants had ever “sent a sexually suggestive nude or nearly nude photo or video of themselves to someone else.” App. 999, 6102. Because of this conflation of images covered and not covered by the Statutes, Zimmermann’s work provides little indication of the quantity of invalid applications of the Statutes to private sexually explicit images. 42 these images would have fallen within the Statutes’ scope. Drouin’s second study was more specific, however, and found that approximately 20% of students in committed or casual sexual relationships had sent “an entirely nude picture or video”16 to their partners, and 10% of students in committed or cheating relationships had sent a picture or video of themselves masturbating. App. 980. A minimal number of cheating and casual sex partners also shared private images of sexual intercourse with another person. Id. Drouin’s studies do not purport to be “representative of all American young adults.” App. 981. Indeed, by focusing on self-selected university students and their sharing habits, these studies represent only a narrow sliver of the potential universe of private sexual imagery created by consenting adults. Nevertheless, although these statistics do not provide a precise picture of just how much private sexual imagery is produced in the United States, they are still informative. They demonstrate that there is some substantial amount of private sexually explicit images that the Statutes burden unnecessarily. Plaintiffs’ evidence of the use of 16 The Government rejects the “entirely nude” statistic because that category could include images of breasts but not genitalia. Although an imperfect measure, this category is still probative of the amount of private images burdened by the Statutes even if it does not provide an exact estimate. 43 Internet communication services for similar purposes also buttresses this finding.17 Accordingly, we agree with Plaintiffs that they have demonstrated the existence of a universe of private sexually explicit images not intended for sale or trade along with, to a limited degree, a universe of sexually explicit images that depict only clearly mature adults. The precise size of these groupings defy easy calculation for the reasons we noted above, but statistical precision has never been required to establish a facial overbreadth claim. For example, in Watchtower Bible & Tract Society of New York, Inc. v. Village of Stratton, 536 U.S. 150 (2002), the Supreme Court struck down as facially overbroad an ordinance making it a misdemeanor to engage in door-to-door advocacy without a permit. In doing so, the Court referenced the “significant amount of spontaneous speech” and the “significant number of persons who support causes anonymously” that the ordinance impacted, but did not suggest that the number of persons engaging in such speech needed to be calculated 17 Plaintiffs also relied on images posted on adult dating and social networking sites to prove the substantiality of private sexually explicit images. Because these images are publicly available, however, we doubt that the Government’s interest is not advanced by application of the Statutes to these images. Similarly, we see no difference between a sexually explicit image produced for artistic, educational, or journalistic purposes and an image produced for more expressly pornographic purposes with respect to whether recordkeeping and identification helps to prevent the exploitation of children during the production of those images. 44 precisely. Id. at 166–67. Requiring exact calculations would convert the overbreadth analysis into a mere numbers game, a proposition we have long rejected. See Aiello, 623 F.2d at 854 (“Merely balancing the number of permissible applications obviously is not sufficient.”). Here, Plaintiffs have proved not only that the problematic applications of the Statute are neither hypothetical nor imaginary, but also that they are not isolated in scope. Further, we do not agree with the District Court that the difficulty of enforcing the Statutes against purely private producers of sexually explicit images counsels in favor of facial validity. In that regard, the District Court cited approvingly law enforcement’s testimony that the Government “has no interest in enforcing the Statutes as to purely private communications and that it would have no conceivable way of even doing this—because it would have no knowledge of those private communications in the first place.” FSC II, 957 F. Supp. 2d at 599–600. As a factual matter, the District Court erred when it accepted that the Government would never be able to enforce the Statutes against private producers. Even if the Government could not target such communications, it is no stretch of the imagination for the Government to become aware of such images inadvertently or through the investigation of other suspected crimes. More fundamentally, as the Supreme Court stated in United States v. Stevens, “the First Amendment protects against the Government; it does not leave us at the mercy of noblesse oblige.” 559 U.S. 460, 480 (2010). If the Statutes’ burdens required us to deem them facially invalid, we would 45 not instead uphold them “merely because the Government promised to use [the Statutes] responsibly.” Id. Although the Government claims it would exercise its prosecutorial discretion appropriately today, those assurances “may one day be modified by the executive branch to permit the exercise of the Statutes’ full authority.” FSC I, 677 F.3d at 539 n.15. Accordingly, “a promise by the government that it will interpret statutory language in a narrow, constitutional manner cannot, without more, save a potentially unconstitutionally overbroad statute.” Id. Nevertheless, the invalid applications of the Statutes that Plaintiffs have demonstrated still pale in comparison with the Statutes’ legitimate applications, which counsels against holding the Statutes facially invalid. Indeed, “[d]eclaring a statute unconstitutional on overbreadth grounds is ‘strong medicine’ and should be used ‘sparingly and only as a last resort.’” FSC I, 677 F.3d at 537 (quoting Broadrick, 413 U.S. at 613). Facial invalidation thus “require[s] that a statute’s overbreadth be substantial, not only in an absolute sense, but also relative to the statute’s plainly legitimate sweep.” Williams, 553 U.S. at 292. This is so because the invalidation of a law that in some of its applications is perfectly constitutional “has obviously harmful effects.” Id. Accordingly, in some circumstances, “the chilling effect of an overbroad law, significant though it may be, cannot justify prohibiting all enforcement of that law—particularly a law that reflects ‘legitimate state interests in maintaining comprehensive controls over harmful, constitutionally unprotected conduct.’” Virginia, 539 U.S. at 119 (quoting Broadrick, 413 U.S. at 615). 46 Such is the case here. Plaintiffs have failed to prove that the invalid applications of the Statutes are substantial relative to the Statutes’ legitimate scope. The Sixth Circuit’s resolution of a similar challenge to the Statutes in Connection Distributing Co. v. Holder, 557 F.3d 321 (6th Cir. 2009) (en banc), is instructive. There the Sixth Circuit found that “the overwhelming majority of applications of § 2257 do not offend the free-speech guarantees of the Constitution, and a ‘vigorous’ enforcement of the ‘substantial overbreadth’ requirement prohibits a party from leveraging a few alleged unconstitutional applications of the statute into a ruling invalidating the law in all of its applications.” Id. at 340. So too here. Plaintiffs present more evidence of impermissible applications of the Statutes in this case than the Sixth Circuit confronted in Connection, but still not enough to deem the Statutes facially overbroad. Given that it is Plaintiffs’ burden to demonstrate substantial overbreadth “from the text of [the law] and from actual fact,” Virginia, 539 U.S. at 122 (alteration in original) (quoting N.Y. State Club Ass’n v. City of N.Y., 487 U.S. 1, 14 (1988)), their First Amendment facial claim fails.18 18 Amici the American Civil Liberties Union of Pennsylvania and the Electronic Frontier Foundation argue that McCutcheon v. FEC, 134 S. Ct. 1434 (2014), requires us to consider whether it would have been difficult for Congress legislatively to exempt the invalid applications of the law that we have identified. Surely if Congress could not have achieved a constitutional purpose through any other means, its efforts would be narrowly tailored. But McCutcheon does not stand for the proposition that the existence of a legislative 47 B. Consideration of the remaining overbreadth factors— “the nature of the activity or conduct sought to be regulated” and “the nature of the state interest underlying the regulation,” FSC I, 677 F.3d at 538 (quoting Gibson, 355 F.3d at 226)—is also central to this conclusion. The nature of the state interest underlying the Statutes—protecting children from sexual exploitation by pornographers—is compelling. “Child pornography harms and debases the most defenseless of our citizens.” Williams, 553 U.S. at 307. “The sexual abuse of a child is a most serious crime and an act repugnant to the moral instincts of a decent people.” Ashcroft v. Free Speech Coal., 535 U.S. 234, 244 (2002). Indeed, “[i]t is alternative that would be less burdensome on speech means Congress is constitutionally required to choose that alternative. Indeed, as the Supreme Court observed, when the Court is not applying strict scrutiny, we still require ‘a fit that is not necessarily perfect, but reasonable; that represents not necessarily the single best disposition but one whose scope is ‘in proportion to the interest served,’ . . . that employs not necessarily the least restrictive means but . . . a means narrowly tailored to achieve the desired objective. Id. at 1456–57 (alterations in original) (quoting Bd. of Trs. of State Univ. of N.Y. v. Fox, 492 U.S. 469, 480 (1989)). Amici’s position is simply a least restrictive means test by another name that is inapplicable under intermediate scrutiny. 48 evident beyond the need for elaboration that a State’s interest in ‘safeguarding the physical and psychological well-being of a minor’ is ‘compelling.’” New York v. Ferber, 458 U.S. 747, 756–57 (1982) (quoting Globe Newspaper Co. v. Superior Court, 457 U.S. 596, 607 (1982)). Thus, few objectives are on par with the “surpassing importance” of the Government’s compelling interest in this case. Id. at 757. Further, the Statutes here represent an effort to quash child pornography after Congress found that direct prohibitions of child pornography had not solved the problem. The Statutes were passed only after Congressional findings “that an extensive interstate market for child pornography continued to exist and that children were still at risk for sexual exploitation by pornographers.” FSC I, 677 F.3d at 535. The financial benefits accruing to producers from using youthful models as well as the financial benefits those models themselves enjoy, together with the difficulty of differentiating youthful adults from minors, all combine to increase the risks of children being exploited. That the Statutes represent an effort to stem the tide of child pornography only after direct prohibitions have been insufficiently effective supports the Statutes’ facial validity. This is not to ignore Plaintiffs’ concerns that much of their expressive conduct is legitimate and protected under the First Amendment. We fully recognize that certain explicit, non-obscene materials have artistic, educational, or other social value. Our resolution of Plaintiffs’ facial challenge is not a value judgment as to Plaintiffs’ work. Indeed, the purely private nature of some of the expressive conduct to which the Statutes apply requires additional sensitivity to the 49 core First Amendment values implicated in this case. See Aiello, 623 F.2d at 854 (“A statutory scheme which encompasses the kind of expressive and associational activity that has been traditionally held to be entitled to a high degree of [F]irst [A]mendment protection should be subjected to closer judicial scrutiny than one which does not.”). Nevertheless, the Statutes’ broad legitimate sweep and the Government’s exceedingly compelling interest in this case counsels against facial overbreadth. Accordingly, Plaintiffs’ First Amendment facial challenge to the Statutes fails. VI. Finally, we address Plaintiffs’ claim that the inspection regime authorized by 28 C.F.R. § 75.5 violates the Fourth Amendment. The Statutes require only that producers “shall make such records available to the Attorney General for inspection at all reasonable times.” 18 U.S.C. § 2257(c); id. § 2257A(c). But 28 C.F.R. § 75.5 authorizes law enforcement—without notice, a warrant, or even suspicion that a violation has occurred—to enter any establishment where a producer’s records under the Statutes are maintained to determine compliance. Plaintiffs focus their Fourth Amendment claim on this regulation’s broad grant of authority. Accordingly, we limit our analysis to this regulation and do not address any Fourth Amendment implications arising from the Statutes directly. A. Before reaching the merits of Plaintiffs’ Fourth Amendment claim, we will address the Government’s 50 justiciability arguments. The Government urges that Plaintiffs lack standing to pursue injunctive relief because they have not demonstrated sufficient threat of injury and their claims of future harm are not redressable through injunctive relief given that no inspection program under the Statutes has been in place since 2008. The Government also points to this lack of an existing inspection regime as proof that Plaintiffs’ Fourth Amendment claims are not ripe. Standing to seek injunctive relief requires a plaintiff to show (1) “that he is under threat of suffering ‘injury in fact’ that is concrete and particularized”; (2) “the threat must be actual and imminent, not conjectural or hypothetical”; (3) “it must be fairly traceable to the challenged action of the defendant”; and (4) “it must be likely that a favorable judicial decision will prevent or redress the injury.” Summers v. Earth Island Inst., 555 U.S. 488, 493 (2009) (citing Friends of Earth, Inc. v Laidlaw Envtl. Servs. (TOC), Inc., 528 U.S. 167, 180–81 (2000)). That some of FSC’s members have previously undergone searches pursuant to the regulations here is not sufficient on its own to confer standing to seek injunctive relief. See Steel Co. v. Citizens for Better Env’t, 523 U.S. 83, 109 (1998) (“‘Past exposure to illegal conduct does not in itself show a present case or controversy regarding injunctive relief . . . if unaccompanied by any continuing, present adverse effects.’” (alteration in original) (quoting O’Shea v. Littleton, 414 U.S. 488, 495–96 (1974))); see also McNair v. Synapse Grp. Inc., 672 F.3d 213, 225 (3d Cir. 2012) (past injuries “may suffice to confer individual standing for monetary relief” but “a plaintiff seeking injunctive relief must demonstrate a likelihood of future 51 harm”). Accordingly, we will focus on the threat of future harm for the purposes of this standing inquiry. Here, despite the lack of an existing inspection regime to implement § 75.5, Plaintiffs are suffering real costs as a condition of compliance with a regulation that they urge is unconstitutional. Sufficient injury exists to confer standing where “the regulation is directed at [Plaintiffs] in particular; it requires them to make significant changes in their everyday business practices; [and] if they fail to observe the . . . rule they are quite clearly exposed to the imposition of strong sanctions,” even where there is no pending prosecution. Pic- A-State Pa., Inc. v. Reno, 76 F.3d 1294, 1300 (3d Cir. 1996) (third alteration in original) (quoting Abbott Labs. v. Gardner, 387 U.S. 136, 154 (1967), abrogated on other grounds, Califano v. Sanders, 430 U.S. 99, 105 (1977)); see also Lozano v. City of Hazleton, 620 F.3d 170, 185 (3d Cir. 2010) (standing exists where plaintiffs were “direct targets of an ordinance they allege to be unconstitutional, complaining of what that ordinance would compel them to do”), vacated on other grounds, 131 S. Ct. 2958 (2011). Here, those Plaintiffs who generate images covered under the Statutes face criminal prosecution if they do not make their records available for at least twenty hours per week as required by regulation. See 18 U.S.C. § 2257(f)(5); id. § 2257A(f)(5); 28 C.F.R § 75.5(c)(1). Even without a formal inspection regime in place, Plaintiffs must still comply with § 75.5’s requirements and be prepared to face an inspection without warning and at law enforcement’s discretion. Each week, Plaintiffs either personally or through a custodian must arrange their businesses to have access to their records during specific 52 times. The cost of complying with this regulation thus affects each producer of sexually explicit images in a concrete way that is sufficient to establish an injury-in-fact. Compounding this injury is that the threat of future inspection is not remote, despite the Government’s assurances to the contrary. There is no dispute that Plaintiffs intend to continue to engage in conduct that subjects them to the Statutes. And nothing prevents law enforcement from resuming inspections pursuant to § 75.5, even if we accept the Government’s representation that it has no current plans to do so. Further, although not sufficient on its own to support standing, the fact that some of FSC’s members have been subjected to records inspections in the past makes future inspections more credible. See Susan B. Anthony List v. Driehaus, 134 S. Ct. 2334, 2345 (2014) (“[P]ast enforcement against the same conduct is good evidence that the threat of enforcement is not ‘chimerical.’” (quoting Steffel v. Thompson, 415 U.S. 452, 459 (1974))). Therefore, we conclude that Plaintiffs have also demonstrated that the threat of future harm is “actual and imminent, not conjectural or hypothetical.” Summers, 555 U.S. at 493. Viewed this way, Plaintiffs’ injury is also redressable. “[S]tanding requires that there be redressability, which is ‘a showing that the injury will be redressed by a favorable decision.’” Constitution Party of Pa. v. Aichele, 757 F.3d 347, 368 (3d Cir. 2014) (internal quotation marks omitted) (quoting Toll Bros. v. Twp. of Readington, 555 F.3d 131, 142 (3d Cir. 2009)). A declaration that § 75.5 is unconstitutional and an injunction barring the Government from conducting searches in the manner currently prescribed would alleviate 53 the costs associated with making records available for physical inspection twenty hours per week and remove the real threat of inspections described above.19 For these 19 The Government does not challenge the traceability requirement, and rightfully so. There is no doubt that the challenged regulation caused the injury-in-fact of which Plaintiffs complain. See Toll Bros., 555 F.3d at 142 (“If the injury-in-fact prong focuses on whether the plaintiff suffered harm, then the traceability prong focuses on who inflicted that harm.”). 54 reasons, we hold Plaintiffs’ Fourth Amendment claims are justiciable.20 B. We turn to the merits of Plaintiffs’ as-applied Fourth Amendment claim.21 In FSC I, we directed the District Court 20 For the same reasons, we hold Plaintiffs’ Fourth Amendment claim is also ripe. Ripeness is a separate doctrine from standing, but both doctrines originate from the same Article III requirement of a case or controversy. Susan B. Anthony List, 134 S. Ct. at 2341 n.5 (citing DaimlerChrysler Corp. v. Cuno, 547 U.S. 332, 335 (2006)); see also Presbytery of N.J. of Orthodox Presbyterian Church v. Florio, 40 F.3d 1454, 1462 (3d Cir. 1994) (standing concerns “who may bring the action” and ripeness involves “when a proper party may bring an action” (emphasis added)). Here, whether Plaintiffs have standing or their claims are ripe for adjudication both turn on whether the threat of future harm under the Statutes is sufficiently immediate to constitute a cognizable injury. See Presbytery of N.J., 40 F.3d at 1462 (“[I]t is of course true that if no injury has occurred, the plaintiff can be told either that she cannot sue, or that she cannot sue yet.” (quoting Smith v. Wis. Dep’t of Agric., Trade & Consumer Prot., 23 F.3d 1134, 1141 (7th Cir. 1994))). Here, the threat of future inspections has caused Plaintiffs to incur ongoing costs to comply with the regulations. Under these circumstances, Plaintiffs’ claims are ripe. 55 to consider whether an inspection under § 75.5 “was a ‘search’ under the Fourth Amendment pursuant to either the reasonable-expectation-of-privacy test set forth in [Katz v. United States, 389 U.S. 347 (1967)] or the common-law- trespass test described in [United States v. Jones, 132 S. Ct. 945 (2012)].” FSC I, 677 F.3d at 544. After developing a thorough record, the District Court concluded that the warrantless inspections conducted pursuant to regulation were searches under both tests. As to the Katz analysis, the inspections invaded areas to which the public did not have access and in which there was a reasonable expectation of privacy (e.g., private offices, storage rooms, and residences). FSC II, 957 F. Supp. 2d at 602–03. And the physical presence of law enforcement officers in those areas also constituted trespasses under the Jones framework. Id. at 603– 04. The Government does not contest this analysis, and we see no reason to reach a different conclusion. The constitutionality of the warrantless searches under the Fourth Amendment thus rises and falls with the administrative search exception to the warrant requirement applicable to closely regulated industries. “Searches conducted absent a warrant are per se unreasonable under the Fourth Amendment, subject to certain exceptions.” United States v. Katzin, 769 F.3d 163, 169 (3d Cir. 2014) (en banc). “[T]he few situations in which a search may be conducted in the absence of a warrant have been carefully delineated and the burden is on those seeking the exemption to show the 21 Because § 75.5 as applied to Plaintiffs violates the Fourth Amendment, we need not, and do not, reach the issue of the regulations’ facial validity. 56 need for it.” California v. Acevedo, 500 U.S. 565, 589 n.5 (1991) (quoting Arkansas v. Sanders, 442 U.S. 753, 759–69 (1979), overruled on other grounds by Acevedo, 500 U.S. 565) (internal quotation marks omitted). But as we explained in FSC I, “[c]ertain industries have such a history of government oversight that no reasonable expectation of privacy could exist.” 677 F.3d at 544. Under these circumstances, “the warrant and probable- cause requirements, which fulfill the traditional Fourth Amendment standard of reasonableness for a government search, have lessened application.” New York v. Burger, 482 U.S. 691, 702 (1987) (citation omitted). Thus, “where the privacy interests of the owner are weakened and the government interests in regulating particular businesses are concomitantly heightened, a warrantless inspection of commercial premises may well be reasonable within the meaning of the Fourth Amendment.” Id. Even if a business is determined to be part of a closely regulated industry, we must then consider whether the warrantless searches themselves are reasonable. This requires examining whether “the following criteria are met: (1) the regulatory scheme furthers a substantial government interest; (2) the warrantless inspections are necessary to further the regulatory scheme; and (3) the inspection program, in terms of certainty and regularity of its application, is a constitutionally adequate substitute for a warrant.” FSC I, 677 F.3d at 544 (citing Burger, 482 U.S. at 702–03). 1. 57 To determine whether an industry is closely regulated, factors to consider include the “duration of the regulation’s existence, pervasiveness of the regulatory scheme, and regularity of the regulation’s application.” Id. Here, the Government points to the fact that since 1978, Congress has criminalized the commercial use of children in sexually explicit materials. See FSC I, 677 F.3d at 525. Since 1988, Congress has imposed recordkeeping requirements similar to those currently embodied in § 2257. Id. Some regulation of sexually explicit images, even those not depicting children, has therefore been in place for some time. But the regulations in this area are not as pervasive as in other industries previously deemed closely regulated. For example, in determining whether the Pennsylvania funeral industry was closely regulated, we looked to the “broad range of standards that funeral directors in Pennsylvania have long been required to comply with,” including licensing requirements, health standards, funeral home services requirements, federal pricing disclosure requirements, and OSHA safety standards. Heffner v. Murphy, 745 F.3d 56, 66 (3d Cir. 2014). Similarly, in finding the New Jersey horse- racing industry closely regulated, we looked to the industry’s licensing requirements for all employees in the industry, prohibitions on employing individuals convicted of certain crimes, and the creation of the New Jersey Racing Commission with broad rulemaking authority. Shoemaker v. Handel, 795 F.2d 1136, 1141 (3d Cir. 1986). The Supreme Court has required a similar degree of pervasive regulation, with “the closely regulated industry . . . [being] the exception,” not the rule. Marshall v. Barlow’s, 58 Inc., 436 U.S. 307, 313 (1978). For example, in New York v. Burger, 482 U.S. 691 (1987), the Court considered whether automobile junkyards were part of a closely regulated industry. In finding them closely regulated, the Court observed that vehicle dismantlers were required to obtain a license, register with the state for a fee, and prominently display a registration number at the junkyard, on business documentation, and on any parts or vehicles passing through the business. Id. at 704. These regulations were backed by civil and criminal penalties. Id. Moreover, junkyards had been regulated for at least 140 years. Id. at 707. In contrast with these industries, the Government fails to identify any similar requirements for producers of sexually explicit images. Nor are the regulations that the Government does identify sufficient. First, the prohibition of child pornography is a broad proscription of a class of images and does not directly target the industry in which Plaintiffs’ are engaged. Nor could it. Plaintiffs’ expression is constitutionally protected, while child pornography is not. See Ferber, 458 U.S. at 764. Indeed, enforcement of the ban is not limited to only those engaged in the business of producing sexually explicit images. The ban on child pornography is therefore more appropriately considered a generally applicable criminal law, not the targeted regulation of any legitimate industry. Although the nature of Plaintiffs’ businesses enhances the chance that they might run afoul of these laws, that alone does not justify deeming the entire industry closely regulated. Second, the Statutes themselves do not justify classifying producers of adult images as closely regulated. 59 To be sure, the Statutes require recordkeeping and labeling. Yet no one is required to obtain a license or register with the Government before producing a sexually explicit image. An artist can pick up a camera and create an image subject to the Statutes without the knowledge of any third party, much less the Government. Nor has the Government identified any regulations governing the manner in which individuals and businesses must produce sexually explicit images. The creation of sexually explicit expression is better characterized by its lack of regulation than by a regime of rules governing such expression. The Government also cannot rely on the regulation’s provision for warrantless searches to itself establish that the industry is closely regulated. The creation of sexually explicit images is not a “new or emerging industr[y]” to which the Government must respond to ensure public health and safety. See Donovan v. Dewey, 452 U.S. 594, 606 (1981) (noting that some new industries, at the time including the nuclear power industry, can be subject to warrantless searches despite “the recent vintage of regulation”). We are doubtful that the Government can create the reduced expectation of privacy of a closely regulated industry to justify warrantless inspections by simply mandating those inspections, particularly where that industry existed long before the regulation’s enactment. See Burger, 482 U.S. at 720 (Brennan, J., dissenting) (“[T]he inspections themselves cannot be cited as proof of pervasive regulation justifying elimination of the warrant requirement; that would be obvious bootstrapping.”). And in any event, as the Government readily acknowledges, no inspections have taken place since 60 2007. This is hardly the “regularity of the regulation’s application,” FSC I, 677 F.3d at 544, that we would expect of a closely regulated industry. For these reasons, we conclude that producers of sexually explicit images are not currently part of a closely regulated industry, and this exception to the warrant requirement does not apply. 2. This alone is sufficient to conclude that the warrantless searches authorized by regulation violate the Fourth Amendment as applied to Plaintiffs. In the interest of completeness, we also address why those inspections are unreasonable, even if producers of sexually explicit images were closely regulated. For this inquiry, we consider whether “(1) the regulatory scheme furthers a substantial government interest; (2) the warrantless inspections are necessary to further the regulatory scheme; and (3) the inspection program, in terms of certainty and regularity of its application, is a constitutionally adequate substitute for a warrant.” FSC I, 677 F.3d at 544 (citing Burger, 482 U.S. at 702–03). We have already discussed the substantiality of the Government’s interest in protecting children with this regulatory scheme, so we need not dwell on the first criterion of this test. And because we find the warrantless inspections here unnecessary, 61 we need not reach whether the inspection program is “a constitutionally adequate substitute for a warrant.”22 Warrantless inspections are necessary where a warrant would undercut the regulatory scheme. But the Government “need not show that warrantless searches are the most necessary way to advance its regulatory interest.” Heffner, 745 F.3d at 68. The need for warrantless searches is most clear where the “administrative inspection scheme[] . . . depend[s] on the element of surprise to both detect and deter violations.” Id. Thus, in Donovan v. Dewey, warrantless inspections to ensure mine safety were necessary because “a warrant requirement could significantly frustrate effective enforcement of the Act” given “the notorious ease with which many safety or health hazards may be concealed if advance warning of inspection is obtained.” 452 U.S. 594, 603 (1981). Similarly, inspections of firearms dealers and junkyards require unannounced, warrantless inspections to prevent the disposal of illicitly held items. Burger, 482 U.S. at 710, 713 (citing United States v. Biswell, 406 U.S. 311, 315 22 The District Court considered these three criteria as factors, as opposed to independent requirements. FSC II, 957 F. Supp. 2d at 605 (describing “three-factor Burger test”). This was error. As the Supreme Court explained, “[a] warrantless inspection, . . . even in the context of a pervasively regulated business, will be deemed to be reasonable only so long as [these] three criteria are met.” Burger, 482 U.S. at 702 (emphasis added). In other words, even if an inspection program is an adequate replacement for a warrant, the Government must still demonstrate that warrantless inspections are necessary in the first instance. 62 (1972)). By contrast, where inspections target conditions that are “relatively difficult to conceal or to correct in a short time,” warrants may be required. Biswell, 406 U.S. at 316 (citing See v. City of Seattle, 387 U.S. 541 (1967)). Here, the Government has all but admitted that warrantless searches are unnecessary. As the District Court found, “[b]oth FBI agents testified that it was highly unlikely that a producer could assemble Section 2257 records” on short notice. FSC II, 957 F. Supp. 2d at 606. And we agree with law enforcement’s testimony that the destruction of evidence is not a real concern, given that to do so would only compound any criminal violation of the Statutes. Further, law enforcement here conducted nearly one third of its inspections under the Statutes after providing notice and without any reports of records fabrication. Thus, the record establishes that the type of records required to be maintained, given their scope as well as the need for indexing and cross- referencing, could not easily be recreated on short notice and violations concealed. Under these circumstances, “inspection warrants could be required and privacy given a measure of protection with little if any threat to the effectiveness of the inspection system.” Biswell, 406 U.S at 316. Because warrantless searches are unnecessary, there is no need to sacrifice even administrative warrants and their accompanying “assurances from a neutral officer that the inspection is reasonable under the Constitution, is authorized by statute, and is pursuant to an administrative plan containing specific neutral criteria.” Marshall, 436 U.S. at 323; see also Martin v. Int’l Matex Tank Terminals-Bayonne, 63 928 F.2d 614, 621 (3d Cir. 1991) (discussing relaxed requirements for administrative warrants).23 VII. For these reasons, we will affirm the District Court’s denial of Plaintiffs’ First Amendment claims, except with regard to 28 C.F.R. § 75.5(c)(1). We will remand to the District Court for further consideration of § 75.5(c)(1)’s constitutionality under the First Amendment. We will also vacate that portion of the District Court’s judgment denying Plaintiffs’ Fourth Amendment claim, and we will remand to the District Court to enter a judgment declaring that the 23 Plaintiffs also urge that the regulations are unconstitutional because they do not provide for pre-enforcement review of law enforcement’s demand to conduct an inspection. Because we conclude warrants are necessary, we need not reach this alternative argument. 64 warrantless searches authorized by 28 C.F.R. § 75.5 as applied to Plaintiffs violate the Fourth Amendment.24 24 Plaintiffs also renew their request for a permanent injunction. The District Court’s denial of a permanent injunction is reviewed for abuse of discretion. eBay Inc. v. MercExchange, L.L.C., 547 U.S. 388, 391 (2006). Relying on the Government’s 2008 disbandment of its inspection program, the District Court held that “Plaintiffs d[id] not face a realistic threat of ‘irreparable harm.’” FSC II, 957 F. Supp. 2d at 609. We note that the existence vel non of a threat of irreparable harm is a different inquiry from whether Plaintiffs have demonstrated injury-in-fact sufficient to support standing, as discussed supra. Because we do not perceive any abuse of discretion and Plaintiffs fail to argue otherwise, we decline to issue a permanent injunction. 65