CHRISTEN, Circuit Judge:

Appellants Project Veritas and Project Veritas Action Fund (collectively, “Project Veritas”) argue that an Oregon statute prohibiting unannounced recordings of oral conversations violates the First Amendment. Project Veritas brings as-applied and facial challenges. It contends that the statute is a content-based restriction on expression that is subject to strict scrutiny and that the statute is facially invalid as overbroad. Because Oregon’s statute does not discriminate on the basis of viewpoint or restrict discussion of an entire topic, we conclude it is content neutral, and that it survives intermediate scrutiny. Because Project Veritas fails to show that any unconstitutional applications of the statute substantially outweigh its constitutional applications, Project Veritas cannot establish facial invalidity. Accordingly, we reject Project Veritas’s claims and affirm the district court’s order dismissing the complaint. I Project Veritas is a nonprofit media organization that engages almost exclusively in undercover journalism. It employs both open and secret audiovisual recording to investigate matters of public concern, sometimes—but not always—in areas open to the public. Whether recording openly or surreptitiously, Project Veritas does not expressly inform individuals that their conversations are being recorded. According to Project Veritas, an announcement that a conversation is being recorded causes individuals to refuse to talk or to distort their story, thereby compromising the quality of Project Veritas’s journalism. Project Veritas maintains that it does not engage in eavesdropping—i.e., the

PROJECT VERITAS V. SCHMIDT 7

interception, without prior consent, of wire or oral communications to which a Project Veritas reporter is not a party. See Or. Rev. Stat. § 165.543(1). Rather, Project Veritas seeks to conduct undercover investigations in Oregon, and it contends that Oregon’s conversational privacy statute prevents it from doing so. See Or. Rev. Stat. § 165.540(1)(c). Section 165.540(1)(c) of the Oregon Revised Statutes requires that notice be given before oral conversations may be recorded. Specifically, the statute provides that “a person may not . . . [o]btain or attempt to obtain the whole or any part of a conversation by means of any device, contrivance, machine, or apparatus, . . . if not all participants in the conversation are specifically informed that their conversation is being obtained.” Id. 1 The statute does not define “specifically informed,” and because prosecutions pursuant to section 165.540(1)(c) are very infrequent, caselaw on this point is sparse at best. 2 A “[c]onversation” is defined as “the transmission between two or more persons of an oral communication which is not a telecommunication or a radio communication, and includes a communication occurring through a video conferencing program.” Or. Rev.

1 A violation of section 165.540(1)(c) is punishable as a misdemeanor. Or. Rev. Stat. § 165.540(9). 2 The dissent cites State v. Haase, 895 P.2d 813 (Or. Ct. App. 1995), for the proposition that section 165.540(1)(c) “bans audiotaping even if the speaker notices that someone has a recording device in her hand.” The dissent misreads Haase. There, the Oregon Court of Appeals reversed an order excluding audio recorded by a police officer’s wearable microphone, concluding that the police officer did not violate section 165.540(1)(c) because the defendant was warned that he was being monitored by audio means and this warning reasonably informed the defendant that the conversation was being obtained. Id. at 815.

8 PROJECT VERITAS V. SCHMIDT

Stat. § 165.535(1). Thus, Oregon’s conversational privacy statute prohibits unannounced audio-only recordings of oral communications between two or more persons, and the audio portion of audiovisual recordings of oral communications. It does not address video-only recordings or photographs. Oregon’s strong interest in protecting conversational privacy dates back to 1955. That year, the Oregon legislature enacted section 165.540(1)(a), a statutory provision that criminalized wiretapping. ^[3] In 1959, the legislature expanded Oregon’s protection of conversational privacy by adding what would become section 165.540(1)(c)—the subsection at issue here—to prohibit the secret “tape recording of face-to-face conversations.” State v. Lissy, 747 P.2d 345, 350 (Or. 1987). Oregon’s general prohibition on unannounced recordings of face-to-face conversations has several exceptions, but Project Veritas focuses its challenge on two of them. The first, the felony exception, allows a person to “record[] a conversation during a felony that endangers human life.” See Or. Rev. Stat. § 165.540(5)(a). The Oregon legislature enacted this exception in 1989, thirty years after passing the general prohibition on unannounced recordings of face-to-face conversations. Or. Rev. Stat. § 165.540(5)(a) (1989). This exception was enacted to “eliminate the requirement that police officers obtain prior court approval before using a ‘body wire’ where felony drug offenses or life-endangering felonies are being committed.”

We next consider whether Oregon’s conversational privacy statute is content based or content neutral. See United States v. Swisher, 811 F.3d 299, 311 (9th Cir. 2016) (en banc). The First Amendment “does not countenance governmental control over the content of messages expressed by private individuals.” Turner Broad. Sys., Inc. v. FCC, 512 U.S. 622, 641 (1994). Thus, laws “that suppress, disadvantage, or impose differential burdens upon speech because of its content” are subject to strict scrutiny. Id. at 642. A regulation that is content neutral, on the other hand, must only satisfy intermediate scrutiny. Id. Project Veritas argues that Oregon’s conversational privacy statute is facially content based because one must examine the content of an unannounced recording to determine whether it is lawful or unlawful. For example, Project Veritas reasons that to determine whether an audio recording is an unlawful unannounced recording of a

10 E.g., Justin Marceau & Alan K. Chen, Free Speech and Democracy in the Video Age, 116 Colum. L. Rev. 991, 997–98 (2016); Joel R. Reidenberg, Privacy in Public, 69 U. Miami L. Rev. 141, 152 (2014); Margot E. Kaminski, Privacy and the Right to Record, 97 B.U. L. Rev. 167, 232–43 (2017).

PROJECT VERITAS V. SCHMIDT 25

conversation in which a Public Records Advocate is a participant, or a lawful unannounced recording of a conversation involving a police officer performing his duties in public, one must listen to the content of the recording. In other words, Project Veritas contends that because the statute imposes limits on what may be recorded, it is necessarily content based. This rigid conception of the content-neutrality inquiry is unsupported by precedent. For decades, the Supreme Court routinely emphasized that “[t]he government’s purpose” in regulating speech “is the controlling consideration” in “determining content neutrality.” Ward v. Rock Against Racism, 491 U.S. 781, 791–92 (1989) (finding content neutral a sound- amplification regulation that applied to an outdoor urban stage); see also Turner, 512 U.S. at 646 (finding content neutral a regulation requiring cable television operators to devote a portion of their channels to local broadcast stations because Congress’s goal “was not to favor programming of a particular subject matter, viewpoint, or format, but rather to preserve access to free television programming”). Put another way, “[t]he principal inquiry in determining content neutrality . . . is whether the government has adopted a regulation of speech because of disagreement with the message it conveys.” Hill v. Colorado, 530 US. 703, 719 (2000) (quoting Ward, 491 U.S. at 791). Thus, “[g]overnment regulation of expressive activity is content neutral so long as it is ‘justified without reference to the content of the regulated speech.’” Ward, 491 U.S. at 791 (quoting Clark, 468 U.S. at 293); see also R.A.V. v. City of St. Paul, 505 U.S. 377, 390 (1992) (approving content discrimination, where speech is proscribable, if “there is no realistic possibility that official suppression of ideas is afoot”).

26 PROJECT VERITAS V. SCHMIDT

Although the Supreme Court has acknowledged that “the mere assertion of a content-neutral purpose” would not “be enough to save a law which, on its face, discriminates based on content,” Turner, 512 U.S. at 642–43, it has not adopted a bright-line rule that any consideration of content mandates strict scrutiny. The Court noted in Hill v. Colorado, a decision issued in 2000, that it had “never held, or suggested, that it is improper to look at the content of an oral or written statement in order to determine whether a rule of law applies to a course of conduct.” See 530 U.S. at 721–22 (rejecting an argument that the examination of oral statements rendered a law content based). In 2015, the Supreme Court decided Reed v. Town of Gilbert, a case concerning a sign code that imposed varying restrictions on different types of signs—ideological signs, political signs, and temporary directional signs. 576 U.S. 155, 159–60 (2015). The code imposed the most stringent restrictions on temporary directional signs. Id. at 160. A local church that had displayed temporary signs to direct people to its upcoming services challenged the statute after it was cited for violating the code’s restrictions. Id. at 161. The Reed Court held that the Town of Gilbert’s sign code was facially content based because the restrictions that applied to a given sign “depend[ed] entirely on the communicative content of the sign.” Id. at 164. “If a sign informs its reader of the time and place a book club will discuss John Locke’s Two Treatises of Government, that sign will be treated differently from a sign expressing the view that one should vote for one of Locke’s followers in an upcoming election, and both signs will be treated differently from a sign expressing an ideological view rooted in Locke’s theory of government.” Id. Although the town may have had a benign purpose in enacting the code, Reed stated that

PROJECT VERITAS V. SCHMIDT 27

courts must “consider[] whether a law is content neutral on its face before turning to the law’s justification or purpose.” Id. at 166. 11 We applied Reed in Animal Legal Defense Fund v. Wasden, where we held that Idaho’s statute prohibiting recordings of the “conduct of an agricultural production facility’s operations” was content based. 878 F.3d at 1203. Wasden concluded that the statute necessarily drew content- based distinctions on its face because one could only “make a determination about criminal liability” by “viewing the recording” of an agricultural production facility’s operations. Id. at 1204. The Supreme Court subsequently undercut this reasoning in City of Austin v. Reagan National Advertising of Austin, LLC, 596 U.S. 61 (2022). In that clarifying decision, the Court considered a First Amendment challenge that centered on a distinction in Austin’s sign code between off-premises signs (i.e., signs that advertise goods or services available at a location separate from where the sign was installed) and on-premises signs (i.e., signs that advertise goods or services available at the location of the sign). Id. at 66. Because the regulatory scheme drew “distinctions based on the message a speaker convey[ed],” Reed, 576 U.S. at 163–64, the plaintiff argued it was content based. City of Austin, 596 U.S. at 69, 74. City of Austin

11 Several Courts of Appeals interpreted Reed’s emphasis on facial neutrality as “a drastic change in First Amendment jurisprudence.” Free Speech Coal., Inc. v. Att’y Gen. U.S., 825 F.3d 149, 160 n.7 (3d Cir. 2016); Reagan Nat’l Advert. of Austin, Inc. v. City of Austin, 972 F.3d 696, 702 (5th Cir. 2020) (describing Reed as a “sea change”), rev’d, 596 U.S. 61 (2022); see also, e.g., Cahaly v. LaRosa, 796 F.3d 399, 405 (4th Cir. 2015); Norton v. City of Springfield, 806 F.3d 411, 412 (7th Cir. 2015).

28 PROJECT VERITAS V. SCHMIDT

firmly rejected the proposition that a “regulation cannot be content neutral if it requires reading the sign at issue,” id. at 69, and emphasized “that restrictions on speech may require some evaluation of the speech and nonetheless remain content neutral,” id. at 72. Ultimately, City of Austin held that Austin’s sign ordinance was content neutral because it “require[d] an examination of speech only in service of drawing neutral, location-based lines.” Id. at 69. To further illustrate this controlling principle, City of Austin relied on earlier Supreme Court cases addressing restrictions on solicitation. “To identify whether speech entails solicitation, one must read or hear it first,” but Supreme Court precedent has long held that “restrictions on solicitation are not content based and do not inherently present ‘the potential for becoming a means of suppressing a particular point of view,’ so long as they do not discriminate based on topic, subject matter, or viewpoint.” Id. at 72 (citation omitted). City of Austin cited the Court’s prior decision in Heffron v. International Society for Krishna Consciousness, Inc., 452 U.S. 640 (1981). See City of Austin, 596 U.S. at 72. There, the Court confronted a Minnesota statute that limited sale, distribution, and solicitation activities at the Minnesota State Fair to a particular location within the fairgrounds. Heffron, 452 U.S. at 643–44. The case arose when Krishna practitioners were prohibited from distributing religious literature and soliciting donations outside the designated location at the fair. Id. at 646, 655. The Court observed that the statute was not “based upon either the content or subject matter of speech,” noting that the statute applied “evenhandedly to all who wish to distribute and sell written materials or to solicit funds.” Id. at 648–49 (citation omitted). Thus, although the statute implicated the content

PROJECT VERITAS V. SCHMIDT 29

of speech—in the sense that its application turned on whether the speech could be characterized as “solicitation”—it did so in a manner that was neutral with respect to the message that individual speakers expressed. Id.; see also City of Austin, 596 U.S. at 73–74 (rejecting the view that examination of speech “inherently triggers heightened First Amendment concern”). The Minnesota statute did not evince an intent to favor or disfavor a particular message or speaker. Further illuminating this threshold principle, City of Austin drew upon Members of City Council of Los Angeles v. Taxpayers for Vincent. See 596 U.S. at 73. There, the Court concluded that an ordinance prohibiting posting signs on public property was content neutral, even though the law included exemptions for, among other things, markers commemorating “historical, cultural, or artistic event[s].” Taxpayers for Vincent, 466 U.S. at 791 n.1, 804, 817. The Court noted that the “general principle” forbidding a government from “favor[ing] some viewpoints or ideas at the expense of others” had “no application” because there was “not even a hint of bias or censorship in the City’s enactment or enforcement of” the ordinance at issue. Id. at 804; see also Renton v. Playtime Theatres, Inc., 475 U.S. 41, 43, 48 (1986) (concluding that a zoning ordinance that placed restrictions on the location of movie theatres based on whether they presented adult content was content neutral because it was not aimed at “suppress[ing] the expression of unpopular views”). These precedents recognize that the “the rationale [for] the general prohibition” on content-based regulations “is that content discrimination raises the specter that the Government may effectively drive certain ideas or viewpoints from the marketplace.” Davenport v. Wash.

30 PROJECT VERITAS V. SCHMIDT

Educ. Ass’n, 551 U.S. 177, 188 (2007) (internal quotation marks omitted) (quoting R.A.V., 505 U.S. at 387). This concern is present not only when a regulation discriminates on the basis of viewpoint, Rosenberger v. Rector & Visitors of Univ. of Va., 515 U.S. 819, 829 (1995), but also when a regulation restricts “discussion of an entire topic,” Consol. Edison Co. of N.Y. v. Pub. Comm’n of N.Y., 447 U.S. 530, 537 (1980). See Reed, 576 U.S. at 168–69; see also id. at 182 (Kagan, J., concurring in the judgment). Critically, City of Austin reaffirmed that this concern is not necessarily implicated by every regulation that depends on the content of protected speech. See City of Austin, 596 U.S. at 73–74. Put another way, not every regulation that turns on the content of speech in the loosest sense is content based in the constitutional sense. A regulation may remain content neutral despite touching on content to distinguish between classes or types of speech—such as speech that constitutes solicitation, Heffron, 452 U.S. at 649–50, or speech that draws neutral, location-based distinctions, City of Austin, 596 U.S. at 69—so long as it does not discriminate on the basis of viewpoint or restrict discussion of an entire topic. See City of Austin, 596 U.S. at 73–74; Reed, 576 U.S. at 168–69. Thus, a regulation that restricts speech involving solicitation, or involving a police officer, may be content neutral. As the Supreme Court has plainly stated, regulations that “confer benefits or impose burdens on speech without reference to the ideas or views expressed are in most instances content neutral.” Turner, 512 U.S. at 643 (emphasis added).

We conclude that section 165.540(1)(c) is content neutral. It places neutral, content-agnostic limits on the PROJECT VERITAS V. SCHMIDT 31

circumstances under which an unannounced recording of a conversation may be made. The rule that emerges from the Supreme Court’s caselaw is that “[a] regulation of speech is facially content based . . . if it ‘target[s] speech based on its communicative content’— that is, if it ‘applies to particular speech because of the topic discussed or the idea or message expressed.’” City of Austin, 596 U.S. at 69 (alteration in original) (quoting Reed, 576 U.S. at 163); see also Turner, 512 U.S. at 643 (stating that “laws that by their terms distinguish favored speech from disfavored speech on the basis of the ideas or views expressed are content based”). Regulations of speech that are facially neutral may nevertheless be content based in their justification if they “cannot be ‘justified without reference to the content of the regulated speech,’” or “were adopted by the government ‘because of disagreement with the message [the speech] conveys.’” Reed, 576 U.S. at 164 (alteration in original) (internal quotation marks omitted) (quoting Ward, 491 U.S. at 791); see also Porter v. Martinez, 68 F.4th 429, 438–39 (9th Cir. 2023). Project Veritas first argues that section 165.540(1)(c), without consideration of its exceptions, is a content-based regulation as applied to Project Veritas’s creation of audio recordings. Project Veritas alleges that because the announcement of a recording “itself alters the content of what will be recorded,” the prohibition on unannounced recordings is content based. For support, Project Veritas relies on precedent addressing government-compelled speech. See Nat’l Inst. of Fam. & Life Advocs. v. Becerra, 585 U.S. 755 (2018); Am. Beverage Ass’n v. City & County of San Francisco, 916 F.3d 749 (9th Cir. 2019) (en banc). This caselaw does not support Project Veritas’s position because Oregon’s conversational privacy statute does not

32 PROJECT VERITAS V. SCHMIDT

“compel[] individuals to speak a particular message.” Becerra, 585 U.S. at 766. Moreover, even though Oregon’s statute limits the circumstances under which a conversation may be recorded, it is facially neutral. That is, it does not “draw[] distinctions based on the message a speaker conveys,” Reed, 576 U.S. at 163, and it was not adopted because of the government’s “disagreement with the [speaker’s] message,” id. at 164 (quoting Ward, 491 U.S. at 791). Project Veritas also argues the statute is content based because it “establishes a general ban on speech, but maintains exceptions for speech on certain subjects.” See Glendale Assocs., Ltd. v. NLRB, 347 F.3d 1145, 1155–56 (9th Cir. 2003). These “certain subjects” are embodied, according to Project Veritas, in the felony exception, Or. Rev. Stat. § 165.540(5)(a), and law enforcement exception, Or. Rev. Stat. § 165.540(5)(b). We start with the felony exception, which applies when a conversation is recorded “during a felony that endangers human life.” § 165.540(5)(a). This exception does not address the content of the audio recording. The plain language of the statute dictates that its application turns on when a recorded conversation occurs, and not the subject matter of that conversation. The conversation need not relate to the felony; indeed, it could encompass any content whatsoever. See also State v. Copeland, 522 P.3d 909, 913 (Or. Ct. App. 2022) (“It makes no difference that the victim also recorded a conversation that did not constitute a felony.”). All that matters is whether a recording occurs “during a felony that endangers human life.” § 165.540(5)(a).

PROJECT VERITAS V. SCHMIDT 33

We reach the same conclusion when considering the law enforcement exception. This exception applies to recordings of conversations “in which a law enforcement officer is a participant,” provided certain other conditions are satisfied. ^[12] § 165.540(5)(b). Like the felony exception, this exception is not content based within the meaning of controlling First Amendment precedent. Instead, as with City of Austin’s sign code ordinance, the law enforcement exception is “agnostic as to [the] content” of a recording. 596 U.S. at 69. It does not concern a “particular viewpoint[]” or prohibit discussion of “an entire topic.” Reed, 576 U.S. at 169 (quoting Consol. Edison, 447 U.S. at 530, 537). 13 The exception in section 165.540(5)(b) applies to conversations that involve law enforcement officers, regardless of what the conversation is about. Put another Oregon argues that it has an important interest in ensuring that its residents know when their conversations are

36 PROJECT VERITAS V. SCHMIDT

being recorded. We easily conclude this is a significant governmental interest. The Supreme Court has expressly acknowledged that “[p]rivacy of communication is an important interest.” Bartnicki v. Vopper, 532 U.S. 514, 532 (2001); see also id. at 518 (describing “fostering private speech” as an interest “of the highest order”). It is also well recognized that protecting this interest “encourage[s] the uninhibited exchange of ideas and information among private parties,” and that “the fear of public disclosure of private conversations might well have a chilling effect on private speech.” Id. at 532–33 (citation omitted); see also Alvarez, 679 F.3d at 605 (concluding that conversational privacy “is easily an important governmental interest”). Project Veritas does not dispute this point as a general matter, but it insists that Oregon’s interest in conversational privacy is effectively limited to preventing eavesdropping— i.e., the situation where one, without consent, intercepts communications to which it is not a party, Or. Rev. Stat. § 165.543(1). According to Project Veritas, if an undercover reporter surreptitiously recording a conversation is a participant in that conversation, the other parties to the conversation have only a minimal privacy interest in not being recorded. To support this assertion, Project Veritas relies on the proposition that the secret recording of one’s speech imposes no greater burden on privacy than having one’s speech heard. We cannot so easily discard Oregon’s interest in conversational privacy. Where one “impart[s] information to strangers, one inevitably risks its secondhand repetition,” but “there is ‘a substantial distinction between the secondhand repetition of the contents of a conversation and

PROJECT VERITAS V. SCHMIDT 37

its simultaneous dissemination to an unannounced second auditor, whether that auditor be a person or a mechanical device.’” Med. Lab’y Mgmt. Consultants v. Am. Broad. Cos., 306 F.3d 806, 815 (9th Cir. 2002) (alteration omitted) (quoting Sanders v. Am. Broad. Cos., 978 P.2d 67, 72 (Cal. 1999)). In a world where one knows that any conversation can be secretly recorded at any time, and subsequently disseminated, it is easy to imagine that there might be a deleterious effect on the “uninhibited exchange of ideas,” and a pervasive “chilling effect on private speech.” Bartnicki, 532 U.S. at 532–33 (citation omitted). 15 As Justice Harlan noted over half a century ago, “[a]uthority is hardly required to support the proposition that words would be measured a good deal more carefully and communication inhibited if one suspected his conversations were being transmitted and transcribed.” United States v. White, 401 U.S. 745, 787 (1971) (Harlan, J., dissenting). If all that is heard may be recorded, such a regime “might well smother that spontaneity—reflected in frivolous, impetuous, sacrilegious, and defiant discourse—that liberates daily life.” Id. Oregon professes an interest in avoiding such a world, and the Supreme Court has identified it as an important one. See Bartnicki, 532 U.S. at 532–33. Acknowledging this distinction between being recorded and merely being heard, other courts have endorsed—in the context of privacy torts—an expectation of limited privacy whereby “a person may reasonably expect privacy against

15 See also Margot E. Kaminski & Shane Witnov, The Conforming Effect: First Amendment Implications of Surveillance, Beyond Chilling Speech, 49 U. Rich. L. Rev. 465, 484 (2014) (“[P]rivacy may be important for the development of new ideas, for challenges to the status quo, for change, and for a vigorous democracy.”); M. Ryan Calo, The Boundaries of Privacy Harm, 86 Ind. L. Rev. 1131, 1159 (2011).

38 PROJECT VERITAS V. SCHMIDT

the electronic recording of a communication, even though he or she had no reasonable expectation as to the confidentiality of the communication’s contents.” Sanders, 978 P.2d at 72 (concluding that although an individual’s voice may be audible to some group of people, the individual may nevertheless reasonably expect his voice to remain secluded from the public at large). 16 These courts have reasoned that privacy “is not a binary, all-or-nothing characteristic,” such that the utterance of a statement to one party precludes any expectation of privacy in that statement; rather, “[t]here are degrees and nuances to societal recognition of our expectations of privacy.” Id. Contrary to Project Veritas’s position, our precedent has long discerned a distinction between merely being heard and being recorded. In our seminal case of Dietemann v. Time, Inc., where reporters for Life Magazine secretly recorded the plaintiff who practiced quack medicine in his home without a license, we considered the plaintiff’s privacy interest in not being secretly recorded. 449 F.2d 245, 249 (9th Cir. 1971) (recognizing the viability of a privacy tort under California law and that the First Amendment “is not a license . . . to intrude by electronic means into the precincts of another’s home or office”). We concluded that although “[o]ne who invites another to his home or office takes a risk that the visitor may not be what he seems, and that the visitor may repeat all he hears and observes when he leaves,” one “does not and should not be required to take the risk that what is heard and seen will be transmitted by photograph or

16 See also, e.g., Jensen v. Sawyers, 130 P.3d 325, 340–41 (Utah 2005); In re Marriage of Tigges, 758 N.W.2d 824, 828 (Iowa 2008); Huskey v. Nat’l Broad. Co., 632 F. Supp. 1282, 1288–89 (N.D. Ill. 1986); but see Med. Lab’y Mgmt. Consultants, 306 F.3d at 815 (concluding that Arizona law would not recognize such a right).

PROJECT VERITAS V. SCHMIDT 39

recording . . . to the public at large.” Id.; see also Theofel v. Farey-Jones, 359 F.3d 1066, 1073 (9th Cir. 2004). A contrary rule, we reasoned, “could have a most pernicious effect upon the dignity of man and it would surely lead to guarded conversations and conduct where candor is most valued.” Dietemann, 449 F.2d at 249. We have no hesitation in concluding that secretly recording a conversation presents privacy concerns that are different in kind, and more corrosive, than merely having one’s oral communications heard and repeated. Recordings are uniquely reliable and powerful methods of preserving and disseminating information, but they are also uniquely reliable and powerful methods of invading privacy. Recordings may be made easily, stored indefinitely, disseminated widely, and played repeatedly. Recordings also may be selectively edited, presented without context, manipulated, and shared across the internet. Because an audio recording device reliably captures the sound that it detects, its usage may also create the illusion of objectivity, even where the recording omits critical context due to selective editing or recording. 17 Thus, the transmission of an accurate recording may nevertheless obscure historical truth. 18

17 See Nancy D. Zeronda, Street Shootings: Covert Photography and Public Privacy, 63 Vand. L. Rev. 1131, 1137 (2010). 18 See also Jeffrey Rosen, The Unwanted Gaze: The Destruction of Privacy in America, 8 (2000) (“Privacy protects us from being misdefined and judged out of context in a world of short attention spans, a world in which information can easily be confused with knowledge.”); Richard A. Epstein, Privacy, Publication, and the First Amendment: The Dangers of First Amendment Exceptionalism, 52 Stan. L. Rev. 1003, 1031 (2000) (“[T]he question of truth is not simply a matter of whether certain isolated statements are true. The question is whether the truth

40 PROJECT VERITAS V. SCHMIDT

Moreover, with the rise of accessible artificial intelligence technologies, anyone can use secret recordings to create convincing audio “deepfakes” in which people appear to say things that they never actually said. 19 With this technology, “the only practical constraint on one’s ability to produce a deepfake [is] access to training materials—that is, audio and video of the person to be modeled.” 20 Oregon’s interest in conversational privacy also extends to ensuring that its residents retain control of their own speech. A party’s “secret monitoring denies the speaker an important aspect of privacy of communication—the right to control the nature and extent of the firsthand dissemination of his statements.” Sanders, 978 P.2d at 72 (citation omitted). A secret recording may enable a party to disseminate another’s oral comments in a way the speaker did not intend. Appropriating another’s speech implicates what the Supreme Court has described as the “principle of autonomy to control one’s own speech.” See Hurley v. Irish- Am. Gay, Lesbian & Bisexual Grp. of Bos., 515 U.S. 557, 574 (1995). “The First Amendment securely protects the freedom to make—or decline to make—one’s own speech,” and “it bears less heavily when speakers assert the right to make other people’s speeches.” Eldred v. Ashcroft, 537 U.S. 186, 221 (2003) (emphasis added) (rejecting a First

counts as a fair and accurate abridgment of the entire record.”); Daniel J. Solove, The Virtues of Knowing Less: Justifying Privacy Protections Against Disclosure, 53 Duke L.J. 967, 1038–39 (2003). 19 Robert Chesney & Danielle Citron, Deepfakes and the New Disinformation War, Foreign Affs. (Dec. 11, 2018), https://www.foreignaffairs.com/articles/world/2018-12-11/deepfakes- and-new-disinformation-war. 20 Id.

PROJECT VERITAS V. SCHMIDT 41

Amendment challenge to the Copyright Term Extension Act); see also Wooley v. Maynard, 430 U.S. 705, 714 (1977) (recognizing the “right to speak freely and the right to refrain from speaking at all”). Oregon’s statute “directly enhance[s] private speech” by allowing individuals to choose not to speak, and thereby protects the “freedom not to speak publicly.” Bartnicki, 532 U.S. at 537 (Breyer, J., concurring) (quoting Harper & Row Publishers, Inc. v. Nation Enters., 471 U.S. 539, 559 (1985)). Project Veritas asserts that most—but not all—of its secret recordings will be made in public places such as cafes and parks. But Oregon’s significant interest in protecting private conversations includes private conversations that occur in public or semi-public locations. There is little doubt that “private talk in public places is common.” Alvarez, 679 F.3d at 606 (citation omitted). “Even in public spaces, people do not expect that their sometimes-sensitive discussions with friends and family members will be available for anyone who wants to record them.” 21 Thus, even if a conversation may be overheard in public, Oregon maintains an interest in preventing its recording. Cf. Katz v. United States, 389 U.S. 347, 351 (1967) (“But what he seeks to preserve as private, even in an area accessible to the public, may be constitutionally protected.”); Carpenter v. United States, 585 U.S. 296, 310 (2018) (“A person does not surrender all Fourth Amendment protection by venturing into the public sphere.”). Oregon’s brief on appeal identifies numerous examples of events that it contends are open to and attended by members of the public, but are still associated with an

21 Marc Jonathan Blitz et al., Regulating Drones Under the First and Fourteenth Amendments, 57 Wm. & Mary L. Rev. 49, 94 (2015).

42 PROJECT VERITAS V. SCHMIDT

expectation that attendees will not make secret recordings, such as twelve-step groups, bible study, and religious services. We agree. For these, and the other reasons identified, we conclude that Oregon has a significant interest in protecting conversational privacy.

The next step is deciding whether Oregon’s statute is narrowly tailored to its significant interest. We conclude that the statute is sufficiently narrow. A regulation “need not be the least speech-restrictive means of advancing” a governmental interest. Turner, 512 U.S. at 662; see also Ward, 491 U.S. at 798. To further its interest in preserving conversational privacy, Oregon adopted a relatively modest notice requirement. Absent an applicable exception, Project Veritas must inform participants in a conversation that they will be recorded before initiating a recording. See Or. Rev. Stat. § 165.540(1)(c). Keeping the purpose of the statute in mind, section 165.540(1)(c) is exceptionally well tailored to protecting Oregonians’ private conversations. By requiring that participants in a conversation be informed before an audio recording begins, but not requiring that they consent to the recording, the statute minimizes the infringement upon Project Veritas’s journalistic efforts while still protecting the interviewees’ right to knowingly participate in Project Veritas’s speech—or not. Once a person is on notice that she will be recorded, she may choose to speak or remain silent. Either way, a noticed recording does not violate a privacy interest. Moreover, consistent with Oregon’s interest in conversational privacy, the statute does not sweep in photography or video recordings; it applies only to recordings of face-to-face oral communications.

PROJECT VERITAS V. SCHMIDT 43

Oregon’s statutory scheme is well tailored because it also accounts for some settings in which people cannot reasonably expect not to have their oral statements recorded. The Oregon legislature crafted several exceptions to account for those situations:

The prohibitions in subsection (1)(c) of this section do not apply to persons who intercept or attempt to intercept oral communications that are part of any of the following proceedings, if the person uses an unconcealed recording device . . . : (A) Public or semipublic meetings such as hearings before governmental or quasi-governmental bodies, trials, press conferences, public speeches, rallies and sporting or other events; (B) Regularly scheduled classes or similar educational activities in public or private institutions; or (C) Private meetings or conferences if all others involved knew or reasonably should have known that the recording was being made.

Or. Rev. Stat. § 165.540(6)(a). These exceptions permit open recordings at public gatherings, including protests, and private meetings in which participants should reasonably expect that they will be recorded. Project Veritas and the dissent argue that these carveouts do not render section 165.540(1)(c) perfectly tailored to Oregon’s stated purpose because the law prohibits recording

44 PROJECT VERITAS V. SCHMIDT

in some situations where participants lack any expectation that their conversation would not be recorded—for example, a loud argument on the street, a political provocateur on a crowded subway, or a drunk, hate-filled conversation in a parking lot. 22 We are not persuaded. The limited question before us is whether Oregon’s conversational privacy statute is sufficiently tailored as a constitutional matter; it is not whether we can conceive of applications of the statute that may appear objectionable if viewed in isolation. Even if fringe examples constitute “conversations” within the meaning of section 165.540(1)(c) and Oregon’s notice requirement is overbroad as applied to them, that does not demonstrate that a “substantial portion of the burden on speech does not serve to advance [Oregon’s] goals.” Ward, 491 U.S. at 799. Moreover, Project Veritas’s resort to these niche examples reinforces the conclusion that the bulk of Oregon’s protection against secret audio recording is targeted at achieving Oregon’s significant interest. 23 Where,

22 Although we consider Project Veritas’s as-applied challenge, the regulation “need not be judged solely by reference to the [conduct] at hand.” Clark, 468 U.S. at 296–97; see also Ward, 491 U.S. at 801 (noting that the validity of a regulation “depends on the relation it bears to the overall problem the government seeks to correct, not on the extent to which it furthers the government’s interests in an individual case”). 23 The dissent erroneously posits that we assume Oregon’s conversational privacy statute does not apply to any scenarios involving “open recordings.” We do not. Rather, we have interpreted the exception embodied in section 165.540(6)(a) to encompass certain open recording activities in which Project Veritas seeks to engage. See supra note 5. The Oregon Court of Appeals’s decision in State v. Bichsel, 790 P.2d 1142 (Or. Ct. App. 1990), cited by the dissent, is not to the contrary. There, decades before Oregon adopted the law enforcement exception, the court ruled that a police officer’s investigatory stop of the defendant in an alley was not a “meeting” within the meaning of section

PROJECT VERITAS V. SCHMIDT 45

as here, a governmental interest “would be achieved less effectively absent the regulation” and the regulation achieves its aim “without . . . significantly restricting a substantial quantity of speech that does not create the same evils,” the regulation is sufficiently narrowly tailored. Id. at 799 & n.7 (citation omitted); see also Galvin v. Hay, 374 F.3d 739, 753 (9th Cir. 2004).

We are also persuaded that section 165.540(1)(c) leaves open ample alternative channels of communication for Project Veritas to engage in investigative journalism and to communicate its message. It is well established that an alternative channel need not be ideal, but merely adequate. See Heffron, 452 U.S. at 654– 55. “[T]he First Amendment does not guarantee the right to communicate one’s views at all times and places or in any manner that may be desired.” Id. at 647. The Supreme Court “generally will not strike down a governmental action for failure to leave open ample alternative channels . . . unless the government enactment will foreclose an entire medium of public expression across the landscape of a particular community or setting.” Menotti v. City of Seattle, 409 F.3d 1113, 1138 (9th Cir. 2005) (citation omitted). A restriction runs afoul of the “alternative channels” requirement if it eliminates the only method of communication by which speakers can convey their message to a particular audience. See, e.g., Bay Area Peace Navy v. United States, 914 F.2d 1224, 1229–30 (9th Cir. 1990). But a regulation does not fail intermediate scrutiny merely

165.540(6)(a) because it was a “mere encounter[].” Id. at 1143, 1144 n.3. Nothing in our opinion conflicts with Bichsel.

46 PROJECT VERITAS V. SCHMIDT

because the other available channels of communication would convey the same message somewhat less conveniently or effectively. See, e.g., Santa Monica Nativity Scenes Comm. v. City of Santa Monica, 784 F.3d 1286, 1298–99 (9th Cir. 2015); One World One Fam. Now v. City & County of Honolulu, 76 F.3d 1009, 1014 (9th Cir. 1996). An alternative channel is adequate if it “permits the more general dissemination of a message.” Frisby v. Schultz, 487 U.S. 474, 483 (1988). We reject Project Veritas’s argument that it will be unable to engage in investigative journalism to communicate its message “across the landscape of a particular community or setting,” Menotti, 409 F.3d at 1138 (citation omitted), or to reach a particular audience, if it cannot secretly record face-to-face conversations. Project Veritas retains numerous alternative channels to engage in its journalistic speech activities. It may employ all the traditional tools of investigative reporting, including talking with sources, reviewing records, taking photographs, recording videos openly during public and semi-public meetings and events, recording videos that do not capture oral conversations, recording conversations after announcing it is doing so, and making use of Oregon’s freedom-of-information laws. See, e.g., Or. Rev. Stat. §§ 165.540(6)(a)(A), 192.311–.431, 192.610–.695. Project Veritas may have its reporters go undercover and report on what they have seen and heard—without secretly recording its targets—as journalists have done for centuries. Powerful exposés authored by people like Nellie Bly,24

24 Diane Bernard, She went undercover to expose an insane asylum’s horrors. Now Nellie Bly is getting her due, Wash. Post (July 18, 2019), https://www.washingtonpost.com/history/2019/07/28/she-went-

PROJECT VERITAS V. SCHMIDT 47

Gloria Steinem, 25 and John Howard Griffin 26 clearly demonstrate what our court has long recognized: “hidden mechanical contrivances” are not “‘indispensable tools’ of newsgathering.” Dietemann, 449 F.2d at 249 (rejecting the argument that the First Amendment accorded journalists immunity from invasion of privacy torts). These many approaches to traditional investigative reporting remain available to Project Veritas and they satisfy the alternative- channels requirement. F Even if we agreed with Project Veritas that the statutory exceptions it challenges are content based, the proper next step would be to consider whether the exceptions may be severed rather than striking down the entirety of Oregon’s conversational privacy statute, as Project Veritas urges us to do. 27

undercover-expose-an-insane-asylums-horrors-now-nellie-bly-is- getting-her-due/. 25 Rachel Chang, Inside Gloria Steinem’s Month as an Undercover Playboy Bunny, Biography.com (Mar. 23, 2020), https://www.biography.com/authors-writers/gloria-steinem-undercover- playboy-bunny. 26 Bruce Watson, Black Like Me, 50 Years Later, Smithsonian Mag. (Oct. 2011), https://www.smithsonianmag.com/arts-culture/black-like-me- 50-years-later-74543463/. 27 If the exceptions were content based, strict scrutiny would apply. Turner, 512 U.S. at 641–42. Under strict scrutiny, a regulation is constitutional only if it “is necessary to serve a compelling state interest” and “is narrowly drawn to achieve that end.” Perry Educ. Ass’n v. Perry Local Educators Ass’n, 460 U.S. 37, 45 (1983). Oregon concedes that it cannot satisfy strict scrutiny. 48 PROJECT VERITAS V. SCHMIDT The severability of a state statute is a matter of state law. Leavitt v. Jane L., 518 U.S. 137, 139 (1996) (per curiam). Under Oregon law, “when one part of a statute is found unconstitutional,” the court should “sever the offending part and save the remainder of the statute, [1] unless the legislature has directed otherwise, [2] unless the parts of the statute are so interconnected that it appears likely that the remaining parts would not have been enacted without the unconstitutional part, or [3] unless the remaining parts are incomplete and cannot be executed in accordance with legislative intent.” Outdoor Media Dimensions, Inc. v. Dep’t of Transp., 132 P.3d 5, 18 (Or. 2006) (applying Or. Rev. Stat. § 174.040). 28 The “legislative preference for severing the offending language and saving the remainder of the statute is conditioned only upon [these] three circumstances.” City Univ. v. State, 885 P.2d 701, 704 (Or. 1994). None of Oregon’s three severability exceptions apply here. The first circumstance identified in Outdoor Media is plainly inapplicable because Oregon’s legislature has not issued any specific direction negating its preference for severability. See § 165.540. As for the second and third Outdoor Media circumstances, there can be no question that the felony exception and the law enforcement exception are not so intertwined with the freestanding conversational privacy statute that it could not survive on its own. In fact, it did exist on its own for decades. Oregon established its conversational privacy statute in 1959 and did not enact the 28 This is consistent with the Supreme Court’s longstanding recognition of a “strong presumption of severability.” Barr v. Am. Ass’n of Pol. Consultants, Inc., 591 U.S. 610, 625 (2020); see also Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 328–31 (2006). PROJECT VERITAS V. SCHMIDT 49 felony exception and law enforcement exception until 1989 and 2015, respectively. Oregon’s general protection against unannounced recordings was operational for thirty years before the modern-day amendments were added to address police officers’ concealed recording devices and over fifty years before a consensus emerged in federal caselaw that the First Amendment protects the right to film police officers in public. See, e.g., Askins, 899 F.3d at 1044. The dissent recognizes that we apply Oregon law to determine severability, cites the correct test from Outdoor Media, and then fails to apply it. Instead, the dissent insists that it is “impossible to sever” the law enforcement exception because, without that exception, the conversational privacy statute would fail to account for the First Amendment right to photograph and record matters of public interest. Our dissenting colleagues’ unstated assumption is that the conversional privacy statute is “so essentially and inseparably connected with and dependent upon the [law enforcement exception] that it is apparent that [it] would not have been enacted without [that exception].” § 174.040(2). But there is no hint that the Oregon legislature would have considered the law enforcement exception so essential that it would have opted to repeal its conversational privacy statute—which existed for over half a century and for several decades after courts began to recognize a right to record matters of public interest, see Fordyce, 55 F.3d at 439—rather than be without the exception. Moreover, the dissent’s view would place Oregon in an insoluble dilemma: If a legislature carves out an exception to accommodate speech that is protected by the First Amendment, as Oregon did here, the dissent would strike down the entire law as unconstitutional; and if a legislature enacts a freestanding prohibition on unannounced recordings, the dissent would 50 PROJECT VERITAS V. SCHMIDT deem the law unconstitutional for failing to recognize First Amendment rights. Even if the exceptions to section 165.540(1)(c) did not survive scrutiny, the appropriate next step would be to sever them, leaving in place the general prohibition on concealed recordings in the applications of section 165.540(1)(c) challenged by Project Veritas. 29 See, e.g., State ex rel. Musa v. Minear, 401 P.2d 36, 39 (Or. 1965) (concluding that an amended statute was invalid and reverting it to its pre- amendment form). G Having concluded that section 165.540(1)(c)—with and without its exceptions—survives intermediate scrutiny as applied to Project Veritas, we next consider Project Veritas’s separate facial overbreadth challenge. In its complaint, Project Veritas seeks a declaration that the statute is 29 Because “a ruling of unconstitutionality frustrates the intent of the elected representatives of the people” and thereby undermines a state’s sovereign interests, federal courts should be exceptionally cautious before striking down a state statute as unconstitutional. Ayotte, 546 U.S. at 329 (alteration accepted) (citation omitted). Accordingly, “the normal rule” is that “partial, rather than facial, invalidation is the required course.” Brockett v. Spokane Arcades, Inc., 472 U.S. 491, 504 (1985). Project Veritas contends that Oregon did not brief severability in the district court and relies on our prior decision in Comite de Jornaleros de Redondo Beach v. City of Redondo Beach, 657 F.3d 936, 951 n.10 (9th Cir. 2011) (en banc), to argue that we may not reach it. Even if some part of Oregon’s conversational privacy statute were constitutionally infirm, we doubt this would be the proper course. Indeed, the Supreme Court has previously faulted our court for failing to consider severability before invalidating an entire state statute. See Brockett, 472 U.S. at 507; see also Ayotte, 546 U.S. at 328–31; New York v. United States, 505 U.S. 144, 186 (1992). Because we conclude that no portion of section 165.540(1)(c) is unconstitutional, we need not resolve this issue here. PROJECT VERITAS V. SCHMIDT 51 unconstitutionally overbroad on its face. Because Project Veritas pursues relief that “reach[es] beyond [its] particular circumstances,” Doe, 561 U.S. at 194, Project Veritas must satisfy the requirements for a facial challenge. In most cases, “a plaintiff cannot succeed on a facial challenge unless he ‘establish[es] that no set of circumstances exists under which the [law] would be valid,’ or he shows that the law lacks a ‘plainly legitimate sweep.’” Moody, 603 U.S. at 723 (alterations in original) (citations omitted). Project Veritas acknowledges that it cannot carry this burden, conceding that the statute may permissibly apply where someone secretly records a private conversation. For instance, in Project Veritas’s view, section 165.540(1)(c) may legitimately apply to recording “a hushed conversation in a secluded hallway, the musings of a friend whispering his life’s woes to another friend, in confidence, in a secluded office, or colleagues discussing confidential medical options in a hospital visitation room.” In our First Amendment analysis, we employ a more lenient though still rigorous standard for facial overbreadth challenges, where “[t]he question is whether ‘a substantial number of [the law’s] applications are unconstitutional judged in relation to the statute’s plainly legitimate sweep.’” Moody, 603 U.S. at 723 (alterations in original) (citation omitted). In other words, “the law’s unconstitutional applications” must “substantially outweigh its constitutional ones.” Id. at 724. “[I]nvalidation for overbreadth is strong medicine that is not to be casually employed.” United States v. Hansen, 599 U.S. 762, 770 (2023) (internal quotation marks and citation omitted). “In the absence of a lopsided ratio, courts must handle unconstitutional applications as they usually do—case-by-case.” Id. The party asserting substantial overbreadth bears the burden of establishing it. 52 PROJECT VERITAS V. SCHMIDT Virginia v. Hicks, 539 U.S. 113, 122 (2003); see also Hansen, 599 U.S. at 784. “The first step in the proper facial analysis is to assess the state laws’ scope.” Moody, 603 U.S. at 724; see also Hansen, 599 U.S. at 770. This step is relatively straightforward here. As we have explained, absent the application of an exception, section 165.540(1)(c) requires one to specifically inform participants in a conversation that they are being recorded. “The next order of business is to decide which of the laws’ applications violate the First Amendment, and to measure them against the rest.” Moody, 603 U.S. at 725. To meet its burden, Project Veritas relies on applications of the statute to its own speech. Throughout its complaint and briefing, Project Veritas’s arguments about the constitutionality of section 165.540(1)(c) focus on conversations in which its reporters will be participants, or oral communications of others that Project Veritas reporters will hear. We do not confine our facial overbreadth analysis to the “heartland applications” alleged by the parties; instead, we must “address the full range of activities” that the statute covers. Id at 724–26. Even putting aside that we have already concluded the statute is constitutional as applied to Project Veritas, these applications represent only a sliver of the conversations to which section 165.540(1)(c) may apply. Project Veritas fails to meet its burden because it makes little effort to identify and weigh the conversational privacy statute’s lawful and unlawful applications, and the conversations it wishes to record are plainly a tiny fraction of the whole. Project Veritas makes passing references to other applications it contends are unconstitutional—e.g., PROJECT VERITAS V. SCHMIDT 53 recording a loud argument on the street or a political provocateur on a subway. The dissent does the same, imagining a public official berating a Chipotle employee or uttering a racial slur on a sidewalk. But even assuming that these examples qualify as face-to-face conversations within the meaning of section 165.540(1)(c) and that the statute is unconstitutional as applied to them, “the ratio of unlawful- to-lawful applications is not lopsided enough to justify the ‘strong medicine’ of facial invalidation for overbreadth.” Hansen, 599 U.S. at 784 (citation omitted); see also Moody, 603 U.S. at 723. We therefore reject Project Veritas’s facial overbreadth challenge and affirm the district court’s judgment. AFFIRMED. BENNETT, Circuit Judge, concurring in the judgment: In 2018, when evaluating the creation of audiovisual recordings, we declared that “the recording process is itself expressive,” meaning that “the creation of audiovisual recordings is speech entitled to First Amendment protection as purely expressive activity.” Animal Legal Def. Fund v. Wasden, 878 F.3d 1184, 1204 (9th Cir. 2018). The breadth of that statement is rooted neither in the history of the Free Speech Clause of the First Amendment nor in any decisions from the Supreme Court. The practical implications of our untethered statement in Wasden are easily seen here. Using Wasden as a jumping-off point, Project Veritas 1 contends that the First Amendment categorically protects purely mechanical activity: pressing an audio record button either 1 I refer to Plaintiffs-Appellants as “Project Veritas.” 54 PROJECT VERITAS V. SCHMIDT in secret or without announcement to record all conversations. The majority holds that “Wasden did not conclude that every act of recording requires expressive decisions, nor that every act of recording implicates the First Amendment.” Maj. at 23. But it stops short of holding that the act of pressing an audio record button either secretly or without announcing to record all conversations is not per se “speech” protected by the First Amendment. For this reason, I write separately. Because there is no historical or precedential foundation to support that simply pressing record in secret or without announcement is always protected speech, I would hold that such an act is not per se “speech” protected by the First Amendment. With that understanding, Project Veritas’s facial challenge fails. 2 I. Project Veritas brings a facial challenge to Oregon Revised Statutes § 165.540, which, with some exceptions, 3 prohibits “[o]btain[ing] or attempt[ing] to obtain the whole or any part of a conversation[ 4] by means of any device, 2 As the majority notes, Project Veritas also brings an as-applied challenge. Maj. at 12. I concur in the majority’s holding that such challenge is ripe. Maj. at 13. I also concur in the majority’s judgment that the as-applied challenge fails for the reasons explained in the majority’s opinion. Maj. at 24–50. 3 There are several exceptions, including for recording a conversation during a felony that endangers human life and for recording conversations with police officers performing their official duties. Or. Rev. Stat. § 165.540(5)(a)–(b). 4 A “‘[c]onversation’ means the transmission between two or more persons of an oral communication which is not a telecommunication or PROJECT VERITAS V. SCHMIDT 55 contrivance, machine or apparatus . . . if not all participants in the conversation are specifically informed that their conversation is being obtained.” Or. Rev. Stat. § 165.540(1)(c). On its face, the provision regulates conduct: a person pressing a record button secretly or openly (but without announcing that he is recording) to record a conversation. The provision applies only to audio recordings, as it prohibits only the obtaining of conversations. See id. § 165.535(1). While there are some exceptions, the statute generally does not distinguish between public and private conversations, and it generally does not matter where the conversations occur. But the statute still allows for audio recording, just not secret or unannounced recording. Project Veritas argues that § 165.540(1)(c) is facially unconstitutional because there is a First Amendment right to press record—secretly or openly without announcing—to capture any conversation. A facial challenge requires us to “determine a law’s full set of applications.” Moody v. NetChoice, LLC, 144 S. Ct. 2383, 2394 (2024). Section 165.540(1)(c) generally prohibits a person from using a recording device to eavesdrop on all private conversations in both public and private locations. 5 This includes, for example, preventing a person from leaving a recording device in a public place to secretly record conversations between others. Under the statute, a person also cannot record in secret, or record without announcement, conversations to which he is a party, unless an exception a radio communication, and includes a communication occurring through a video conferencing program.” Or. Rev. Stat. § 165.535(1). 5 Project Veritas clarified at oral argument that it does not challenge the prohibition on eavesdropping. Oral Argument at 15:55–16:04. 56 PROJECT VERITAS V. SCHMIDT applies. Nor may a person openly record all conversations between others without announcing that he is recording, unless, again, an exception applies. A. “The First Amendment, applicable to the States through the Fourteenth Amendment, prohibits the enactment of laws ‘abridging the freedom of speech.’” Reed v. Town of Gilbert, 576 U.S. 155, 163 (2015) (quoting U.S. Const. amend. I). The Free Speech Clause went through several iterations before its adoption as part of the First Amendment. The initial version of the Free Speech Clause as drafted by James Madison was much more specific: “The people shall not be deprived or abridged of their right to speak, to write, or to publish their sentiments; and the freedom of the press, as one of the great bulwarks of liberty, shall be inviolable.” 1 Annals of Cong. 451 (1789) (Joseph Gales ed., 1834). A special committee of the House of Representatives rewrote the Free Speech Clause to read: “The freedom of speech and of the press, and the right of the people peaceably to assemble and consult for their common good, and to apply to the Government for redress of grievances, shall not be infringed.” Id. at 759. The Senate then finalized the language of the First Amendment. Cong. Rsch. Serv., The Constitution of the United States of America: Analysis and Interpretation 1396 (2023). The Free Speech Clause, however, received little debate in the House and there is no record of debate over the clause in the Senate. Id. at 1397 & n.5. Ultimately, the Founders elected to enumerate “simple, acknowledged principles” because “the ratification [of specific abstract propositions would] meet with but little difficulty.” 1 Annals of Cong. 766 (1789). The freedom of the press “was everywhere a PROJECT VERITAS V. SCHMIDT 57 grand topic for declaration, but the insistent demand for its protection on parchment was not accompanied by a reasoned analysis of what it meant, how far it extended, and under what circumstances it might be limited.” Leonard W. Levy, Legacy of Suppression 214–15 (1960). The lack of specificity in the First Amendment almost immediately led to controversies about what was and was not permitted by the Free Speech Clause. One such controversy followed Congress’s passage of the Sedition Act of 1798, which made it a crime for “any person [to] write, print, utter or publish . . . any false, scandalous and malicious writing or writings against the government of the United States, or either house of the Congress . . . , or the President . . . , with intent to defame . . . or to bring them . . . into contempt.” Ch. 74, 1 Stat. 596. Even the Founders were divided on whether the law was constitutional under the First Amendment, with Thomas Jefferson condemning the Act and President John Adams using it to prosecute his political opponents. Erwin Chemerinsky, Constitutional Law: Principles and Policies 1006 (7th ed. 2023). The Sedition Act expired in 1801 before its constitutionality could be challenged in the Supreme Court, 6 but the fact that even the drafters of the Constitution debated the bounds of the Free Speech Clause shows that the “simple, acknowledged principles” underlying the Clause are not so easily put into practice. The Supreme Court has made clear that it “reject[s] the view that freedom of speech . . . , as protected by the First 6 In New York Times Co. v. Sullivan, 376 U.S. 254 (1964), the Supreme Court noted that “[a]lthough the Sedition Act was never tested in this Court, the attack upon its validity has carried the day in the court of history.” Id. at 276 & n.16 (footnote omitted). 58 PROJECT VERITAS V. SCHMIDT and Fourteenth Amendments, are ‘absolutes.’” Konigsberg v. State Bar of Cal., 366 U.S. 36, 49 (1961); see Stromberg v. California, 283 U.S. 359, 368 (1931) (“The right [of free speech] is not an absolute one, and the State in the exercise of its police power may punish the abuse of this freedom.”). “Line-drawing is inevitable as to what speech will be protected under the First Amendment . . . . Lines also must be drawn in defining what is speech.” Chemerinsky, supra, at 1007. That line drawing is even more difficult when, as here, what is sought to be protected is not speech in the traditional sense but conduct. “It is possible to find some kernel of expression in almost every activity a person undertakes—for example, walking down the street or meeting one’s friends at a shopping mall—but such a kernel is not sufficient to bring the activity within the protection of the First Amendment.” City of Dallas v. Stanglin, 490 U.S. 19, 25 (1989). When an individual does not express views through printed or spoken words it is necessary “to determine whether his activity was sufficiently imbued with elements of communication to fall within the scope of the First and Fourteenth Amendments.” Spence v. Washington, 418 U.S. 405, 409 (1974) (per curiam). The Supreme Court asks whether non-speech conduct carries elements of communication because the Court “cannot accept the view that an apparently limitless variety of conduct can be labeled ‘speech’ whenever the person engaging in the conduct intends thereby to express an idea.” United States v. O’Brien, 391 U.S. 367, 376 (1968). That inquiry depends on “the nature of [the individual’s] activity, combined with the factual context and environment in which it was undertaken.” Spence, 418 U.S. at 409–10. For instance, conduct that has involved forms of symbolism, such as conduct involving flags, consistently has been found PROJECT VERITAS V. SCHMIDT 59 to be expressive and protected by the First Amendment. See, e.g., Stromberg, 283 U.S. at 369–70 (striking down a law that prohibited the display of a flag as a symbol of opposition to organized government, because that statute “embrac[ed] conduct which the State could not constitutionally prohibit,” such as “permit[ting] the punishment of the fair use of [an] opportunity” to display a sign and protest the government); W. Va. State Bd. of Educ. v. Barnette, 319 U.S. 624, 632 (1943) (invalidating a law requiring students to salute the American flag and noting “[s]ymbolism is a primitive but effective way of communicating ideas”). Additionally, the context “for purposes of expression is important, for the context may give meaning to the symbol” or conduct. Spence, 418 U.S. at 410. The wearing of black armbands in school could convey a clear message about an issue of public concern, like the Vietnam War, and be subject to First Amendment protections. Tinker v. Des Moines Indep. Cmty. Sch. Dist., 393 U.S. 503, 514 (1969). Affixing a peace sign to an upside-down flag “at a time of national turmoil,” such as after the killing of students at Kent State University, also rises to the level of expressive conduct amounting to speech under the First Amendment. Spence, 418 U.S. at 410, 414 n.10. In both instances, there was “[a]n intent to convey a particularized message,” and “in the surrounding circumstances the likelihood was great that the message would be understood by those who viewed it.” Id. at 410–11. The text of the First Amendment is silent on what exactly is meant by “speech.” The history of the Free Speech Clause offers little guidance, and the Founders themselves had different understandings of what was prohibited by the Clause. But in drawing the bounds of “speech,” the Supreme Court has differentiated between conduct intended to convey 60 PROJECT VERITAS V. SCHMIDT a message and expressionless conduct. 7 Pressing a record button, whether in secret or without announcement, is purely mechanical. That conduct conveys no message. Thus, it is not per se “speech” under the First Amendment. ^[8] B. Even when conduct relates to speech, Supreme Court cases counsel that that conduct may not be speech protected by the First Amendment, particularly when the regulation of such conduct still permits the speaker to express his desired message and there are important countervailing interests. For example, in Zemel v. Rusk, 381 U.S. 1 (1965), the appellant claimed that the Secretary of State’s “refusal to validate his passport for travel to Cuba,” because of the United States’s breaking of diplomatic ties with Cuba and implementation of a travel ban, “denie[d] him rights guaranteed by the First Amendment.” Id. at 16. This was allegedly so because the travel ban “direct[ly] interfere[d] with the First Amendment rights of citizens to travel abroad so that they might acquaint themselves at first hand with the effects abroad of our Government’s policies, foreign and domestic, and with conditions abroad which might affect such policies.” Id. Although the Court acknowledged that 7 Even for expressive conduct, the Supreme Court has clarified that conduct is not immune from government regulation simply because it might communicate a message. O’Brien, 391 U.S. at 376 (“We cannot accept the view that an apparently limitless variety of conduct can be labeled ‘speech’ whenever the person engaging in the conduct intends thereby to express an idea.”). 8 And as explained below, even if some button pushing could amount to protected speech, Project Veritas’s facial challenge would still fail given that the default should be that secret and unannounced recordings are not per se protected under the First Amendment. PROJECT VERITAS V. SCHMIDT 61 the ban did burden the free flow of information, it did not implicate the First Amendment because the ban was ultimately “an inhibition of action.” Id. The Court observed: There are few restrictions on action which could not be clothed by ingenious argument in the garb of decreased data flow. For example, the prohibition of unauthorized entry into the White House diminishes the citizen’s opportunities to gather information he might find relevant to his opinion of the way the country is being run, but that does not make entry into the White House a First Amendment right. The right to speak and publish does not carry with it the unrestrained right to gather information. Id. at 16–17 (emphasis added). In Branzburg v. Hayes, 408 U.S. 665 (1972), the Court relied on Zemel and held that requiring journalists to testify before grand juries did not violate the First Amendment— even though “news gathering may be hampered” by the requirement. Id. at 684; see also id. at 684 n.22 (quoting Zemel’s proposition that there must be limits on the types of conduct protected by the First Amendment). And though the Court acknowledged that requiring journalists to testify would burden some of their newsgathering, the Court nevertheless found that the requirement imposed no burden on their protected speech: [T]hese cases involve no intrusions upon speech or assembly, no prior restraint or restriction on what the press may publish, and 62 PROJECT VERITAS V. SCHMIDT no express or implied command that the press publish what it prefers to withhold. No exaction or tax for the privilege of publishing, and no penalty, civil or criminal, related to the content of published material is at issue here. The use of confidential sources by the press is not forbidden or restricted; reporters remain free to seek news from any source by means within the law. No attempt is made to require the press to publish its sources of information or indiscriminately to disclose them on request. Id. at 681–82. The Branzburg Court explained that sometimes conduct, even when it might further speech, must be regulated to protect important countervailing interests: Although stealing documents or private wiretapping could provide newsworthy information, neither reporter nor source is immune from conviction for such conduct, whatever the impact on the flow of news. Neither is immune, on First Amendment grounds, from testifying against the other, before the grand jury or at a criminal trial. The Amendment does not reach so far as to override the interest of the public in ensuring that neither reporter nor source is invading the rights of other citizens through PROJECT VERITAS V. SCHMIDT 63 reprehensible conduct forbidden to all other persons. Id. at 691–92. Branzburg reiterated Zemel’s holding that there must be limits on the types of conduct protected by the First Amendment. It established that the regulation of some conduct—even when it may impact speech—simply does not implicate the First Amendment, particularly when the speaker is still allowed to express his desired message and the regulation is needed to protect important countervailing interests. In two cases following Branzburg, the Court again confirmed that some speech-related conducted is not protected by the First Amendment. In Pell v. Procunier, 417 U.S. 817 (1974), journalists challenged a California Department of Corrections regulation that “prohibit[ed] face-to-face interviews between press representatives and individual inmates whom they specifically name and request to interview.” Id. at 819. Even though this regulation “clearly restrict[ed] one manner of communication,” id. at 823, the Court held that it did not violate the First Amendment, id. at 835. The Court focused on considerations similar to those identified in Branzburg: there must be limits on protected conduct under the First Amendment, id. at 834 n.9 (quoting Zemel, 381 U.S. at 16– 17); this was the “sole limitation on newsgathering in California prisons,” id. at 831; and the regulation was adopted to mitigate “disciplinary problems caused, in part, by [the department’s prior] liberal posture with regard to press interviews,” id. at 832. 64 PROJECT VERITAS V. SCHMIDT Saxbe v. Washington Post Co., 417 U.S. 843 (1974), involved a similar regulation that “prohibited any personal interviews between newsmen and individually designated federal prison inmates.” Id. at 844. The Saxbe Court found the case “constitutionally indistinguishable” from Pell and thus held that the regulation did not violate the First Amendment. Id. at 850. These cases support that all secret or unannounced audio recordings cannot be per se protected speech under the First Amendment, even if some of those acts of secret or unannounced recording could be indirectly linked to speech. First, we know that, even if the act of recording might be related to speech, the act itself does not automatically qualify as protected speech. There must be limits. See Zemel, 381 U.S. at 16–17. We have expressed this sentiment. Dietemann v. Time, Inc., 449 F.2d 245, 249 (9th Cir. 1971) (“The First Amendment is not a license to trespass, to steal, or to intrude by electronic means into the precincts of another’s home or office.”). Second, a prohibition on secret and unannounced audio recording permits a speaker to use other means of “capturing” the audio contents of a conversation. He can still record, so long as he announces that he is doing so. And, without any announcement, he can still write or type notes during the conversation; he can still write or type notes immediately after the conversation; and he can still dictate the contents of the conversation using a recording device after the conversation. The prohibition on secret and unannounced audio recording also does not restrict his ability to communicate the information that he obtained from the conversation. PROJECT VERITAS V. SCHMIDT 65 Finally, there is a strong countervailing interest protected by the regulation of secret or unannounced audio recording: the interest in maintaining the privacy of communication. “Privacy of communication is an important interest,” as it “encourage[es] the uninhibited exchange of ideas and information among private parties.” Bartnicki v. Vopper, 532 U.S. 514, 532 (2001) (citation omitted). Indeed, “[t]here is necessarily, and within suitably defined areas, a concomitant freedom not to speak publicly, one which serves the same ultimate end as freedom of speech in its affirmative aspect.” Id. at 532 n.20 (quoting Harper & Row, Publishers, Inc. v. Nation Enters., 471 U.S. 539, 559 (1985)). The Court has acknowledged: In a democratic society privacy of communication is essential if citizens are to think and act creatively and constructively. Fear or suspicion that one’s speech is being monitored by a stranger, even without the reality of such activity, can have a seriously inhibiting effect upon the willingness to voice critical and constructive ideas. Id. at 533 (quoting President’s Comm’n on L. Enf’t & Admin. of Just., The Challenge of Crime in a Free Society 202 (1967)). Permitting a person to audio record in secret or record without announcing all conversations—word for word—would inhibit the free exchange of ideas and information, particularly given the reality that audio recordings can be instantly broadcast to the world using the internet. Given these considerations, the secret or unannounced audio recording of all conversations is not per se protected 66 PROJECT VERITAS V. SCHMIDT speech under the First Amendment. Thus, neither the text or the history of the First Amendment, nor Supreme Court precedent interpreting the Free Speech Clause, supports that the act of pressing an audio record button to record all conversations—either in secret or without announcement— is per se speech protected by the First Amendment. Our precedent also offers no persuasive reason to conclude otherwise. C. In Wasden, we stated that “the recording process is itself expressive” and “the creation of audiovisual recordings is speech entitled to First Amendment protection.” 878 F.3d at 1204. In making those broad statements, we primarily relied on Anderson v. City of Hermosa Beach, 621 F.3d 1051 (9th Cir. 2010). See Wasden, 878 F.3d at 1203–04. But Anderson never addressed whether all recordings—let alone secret and unannounced audio recordings—are protected speech. Anderson involved tattooing. 621 F.3d at 1055. There, we held that “the process of tattooing is purely expressive activity” protected by the First Amendment. Id. at 1061–62. In reaching that conclusion, we likened the process of tattooing to “the processes of writing words down on paper, painting a picture, and playing an instrument,” which “are purely expressive activities entitled to full First Amendment protection.” Id. at 1062. And we emphasized that tattoos, which could not be divorced from the process of tattooing itself, involves “skill, artistry, and care”—expressive elements. Id. at 1061. Tattooing, writing, painting, and producing music by playing an instrument all convey messages and are thus all forms of expressive conduct. Id. at 1062. Anderson, however, never considered whether all PROJECT VERITAS V. SCHMIDT 67 audio recordings, even those lacking expressive elements because the recorder simply pushes a button and nothing more, are speech protected by the First Amendment. And pushing a record button is different from the activities Anderson identifies as expressive: unlike painting or writing, pressing a record button requires no such “skill, artistry, and care.” Id. at 1061. Thus, Anderson does not support Wasden’s broad statements. Indeed, in addressing recordings, our sister circuits have not gone so far as holding that all recordings are protected speech. In People for the Ethical Treatment of Animals, Inc. v. North Carolina Farm Bureau Federation, Inc., 60 F.4th 815 (4th Cir. 2023) (“PETA”), the Fourth Circuit expressly declined to adopt the view that all recordings are protected: Our main point of disagreement centers around the [district] court’s belief that all “recording is protected speech.” We do not think it wise to go that far where the case itself does not call for a categorical pronouncement and where the briefing is, understandably, agnostic on the potential implications of such an absolute decision. Should posting a hidden camera in a CEO’s office—or her home—per se constitute protected expression? How about photographing proprietary documents to tap into trade secrets, with no intent of creating a work of art? Recording private telephone conversations? Id. at 836 (citation omitted). 68 PROJECT VERITAS V. SCHMIDT The Fourth Circuit explicitly declined to follow Wasden, questioning Wasden’s “expansive ruling.” Id. at 837 n.9 (“Nor has the Ninth Circuit been able to stress-test the outer limits of its expansive ruling—Wasden itself concerned only recordings of ‘the conduct of an agricultural production facility’s operations.’” (quoting Wasden, 878 F.3d at 1204)). Instead, the PETA court followed other circuits that had declined to hold that all recordings are protected. Id. at 836– 37. For example, PETA explained that the Third Circuit had “prudently declined to ‘say that all recording is protected or desirable.’” Id. at 836 (quoting Fields v. City of Philadelphia, 862 F.3d 353, 360 (3d Cir. 2017)). PETA also noted that the Tenth and Seventh Circuits had taken similar “circumscribed” approaches. Id. at 836–37 (discussing W. Watersheds Project v. Michael, 869 F.3d 1189 (10th Cir. 2017), and ACLU of Ill. v. Alvarez, 679 F.3d 583 (7th Cir. 2012)). Even the dissenting opinion by Judge Rushing expressed that the PETA majority had “rightly reject[ed]” Wasden because “the mere act of recording by itself is not categorically protected speech.” Id. at 845 (Rushing, J., dissenting). In short, our precedent provides no persuasive basis for finding that all audio recordings, including secret and unannounced ones, are categorically protected speech. ^[9] No 9 The majority discusses two other cases from our court that involved recordings. Maj. at 21. But neither held that all audio recording, including secret and unannounced audio recording, is per se speech. See Fordyce v. City of Seattle, 55 F.3d 436, 439 (9th Cir. 1995) (recognizing a “First Amendment right to film matters of public interest”); Askins v. U.S. Dep’t of Homeland Sec., 899 F.3d 1035, 1044 (9th Cir. 2018) (recognizing that the “First Amendment protects the right to photograph and record matters of public interest”). PROJECT VERITAS V. SCHMIDT 69 other circuit has come to that conclusion. 10 Beyond Wasden, the parties point to no cases holding that the mere act of pushing record is categorically protected speech. And, as discussed above, while no Supreme Court case has dealt with secret or unannounced audio recordings, Supreme Court cases addressing the regulation of conduct do not support finding that all such recordings are protected speech under the First Amendment. The majority notes that it “do[es] not 10 The majority states that “[t]he First, Third, Fourth, Fifth, Seventh, Eighth, Tenth, Eleventh, and D.C. Circuits have recognized that many acts of recording qualify as speech and are entitled to First Amendment protection,” Maj. at 21 n.8, but none of the cited cases established a categorical rule that all audio recording, including secret and unannounced audio recording, is speech. See Alvarez, 679 F.3d at 586 (addressing only an as-applied challenge to the open “audiovisual recordings of police officers performing their duties in public places and speaking at a volume audible to bystanders”); Fields, 862 F.3d at 359– 60 (declining to find that the “act of recording” is “inherently expressive conduct” like making art, id. at 359, and explaining that “[w]e do not say that all recording is protected or desirable,” id. at 360); PETA, 60 F.4th at 836 (declining to hold that “all ‘recording is protected speech’” (citation omitted)); W. Watersheds, 869 F.3d at 1197 (“This is not to say that all regulations incidentally restricting access to information trigger First Amendment analysis.”); Turner v. Lieutenant Driver, 848 F.3d 678, 688–90 (5th Cir. 2017) (concluding that “the First Amendment protects the act of making film”); Glik v. Cunniffe, 655 F.3d 78, 83–84 (1st Cir. 2011) (concluding that “the First Amendment protects the filming of government officials in public spaces,” id. at 83, but “the right to film is not without limitations,” id. at 84); Smith v. City of Cumming, 212 F.3d 1332, 1333 (11th Cir. 2000) (concluding that the First Amendment protects the right to visually “record matters of public interest”); Ness v. City of Bloomington, 11 F.4th 914, 923 (8th Cir. 2021) (concluding that taking photographs and recording videos to document alleged noncompliant uses of a public park are entitled to First Amendment protection); Price v. Garland, 45 F.4th 1059, 1071 (D.C. Cir. 2022) (“[P]rohibiting the recording of a public official performing a public duty on public property is unreasonable . . . .”). 70 PROJECT VERITAS V. SCHMIDT suggest that any conduct related in some way to speech creation, however attenuated, is necessarily entitled to First Amendment protection.” Maj. at 18–19. I do not disagree, but I would go further and hold that, when the conduct is purely mechanical, and by itself neither conveys nor contains a message, it is categorically not speech. There should be no general presumption that pushing a button is itself speech. * * * Based on the above, I would hold that the act of pressing a record button secretly or openly but without announcement that one is recording is by itself not categorically protected speech under the First Amendment. Under that premise, Project Veritas’s facial challenge fails because, even assuming that there might be some circumstances when secret or unannounced audio recordings could be protected, those unconstitutional applications do not substantially outweigh the constitutional ones. II. In United States v. Hansen, 599 U.S. 762 (2023), the Supreme Court set forth a demanding standard for evaluating whether a statute is facially overbroad in violation of the First Amendment. Id. at 769–70. Hansen had argued that a federal law prohibiting “encourag[ing] or induc[ing]” illegal immigration was facially overbroad under the First Amendment. Id. at 766. To address Hansen’s argument, the Court stated that “[t]o justify facial invalidation, a law’s unconstitutional applications must be realistic, not fanciful, and their number must be substantially disproportionate to the statute’s lawful sweep.” Id. at 770 (emphasis added). “In the absence of a lopsided ratio, courts must handle PROJECT VERITAS V. SCHMIDT 71 unconstitutional applications as they usually do—case-by- case.” Id. (emphasis added). In applying that standard, the Court first determined what conduct is covered by the statute, i.e., whether “encourage” and “induce” as used in the statute refer to “criminal solicitation and facilitation (thus capturing only a narrow band of speech) or instead as those terms are used in everyday conversation (thus encompassing a broader swath).” Id. at 770–71. Because the Court held that the statute “reaches no further than the purposeful solicitation and facilitation of specific acts known to violate federal law,” it “does not ‘prohibi[t] a substantial amount of protected speech’ relative to its ‘plainly legitimate sweep.’” Id. at 781 (alteration in original) (quoting United States v. Williams, 553 U.S. 285, 292 (2008)). The Court then compared the statute’s constitutional applications to its unconstitutional ones. On the one hand, the statute “encompasses a great deal of nonexpressive conduct— which does not implicate the First Amendment at all.” Id. at 782 (citing as examples “smuggling noncitizens into the country, providing counterfeit immigration documents, and issuing fraudulent Social Security numbers to noncitizens” (citations omitted)). On the other hand, “Hansen fail[ed] to identify a single prosecution for ostensibly protected expression in the 70 years since Congress enacted [the statute]’s immediate predecessor.” Id. Thus, the Court concluded that Hansen failed to show that the statute’s overbreadth is “‘substantial . . . relative to [its] plainly legitimate sweep.’” Id. at 784 (alterations in original) (quoting Williams, 553 U.S. at 292). The Court further reasoned that “[e]ven assuming that [the statute] reaches some protected speech, and even assuming that its application to all of that speech is unconstitutional, the ratio 72 PROJECT VERITAS V. SCHMIDT of unlawful-to-lawful applications is not lopsided enough to justify the ‘strong medicine’ of facial invalidation for overbreadth.” Id. (quoting Broadrick v. Oklahoma, 413 U.S. 601, 613 (1973)). As discussed, I would find that the act of recording conversations, in secret or without announcement, is not per se protected “speech” under the First Amendment. With this understanding, and applying Hansen’s demanding test, I agree with the majority that Project Veritas’s facial challenge fails. Even assuming that Project Veritas has identified some unlawful applications of § 165.540(1)(c), such “applications represent only a sliver of the conversations to which § 165.540(1)(c) may apply.” Maj. at 52. Indeed, they cannot be “substantially disproportionate to the statute’s lawful sweep,” Hansen, 599 U.S. at 770, given that the default should be that secret and unannounced recordings are not per se protected under the First Amendment. Thus, as in Hansen, “[e]ven assuming that [§ 165.540(1)(c)] reaches some protected speech, and even assuming that its application to all of that speech is unconstitutional, the ratio of unlawful-to-lawful applications is not lopsided enough to justify the ‘strong medicine’ of facial invalidation for overbreadth.” Id. at 784 (quoting Broadrick, 413 U.S. at 613). For this reason, Project Veritas’s facial challenge fails. III. I concur that Project Veritas’s facial challenge fails. But I believe that it fails because the act of pressing an audio record button either in secret or without announcement to record all conversations is not per se “speech” protected by the First Amendment. PROJECT VERITAS V. SCHMIDT 73 LEE, Circuit Judge, dissenting, with whom COLLINS, Circuit Judge, joins. Journalists, as well as regular citizens, routinely record the powerful and the privileged behaving badly. Today’s decision imperils the right to capture such abuses of power and other newsworthy events. Oregon does not just ban surreptitious recordings that may implicate privacy concerns: It also criminalizes audio- recording someone—even conversations in public with no reasonable expectation of privacy—if “not all participants in the conversation are specifically informed that their conversation is being obtained.” Or. Rev. Stat. § 165.540(1)(c). So, for example, a citizen in Oregon cannot lawfully audiotape a public official berating an employee at a Chipotle or uttering a racial slur on a public sidewalk— unless that citizen expressly tells that official he is being recorded. We have held that the First Amendment protects the act of recording as an “inherently expressive” activity. Animal Legal Def. Fund v. Wasden, 878 F.3d 1184, 1203 (9th Cir. 2018). The majority opinion, however, upholds the Oregon law under intermediate scrutiny, ruling that the law is narrowly tailored to further the government’s important interest in conversational privacy. I respectfully dissent because Oregon’s law is grossly overbroad and not narrowly tailored to advance the state’s interest in conversational privacy (even assuming intermediate scrutiny applies). Oregon prevents citizens from recording even in public areas if they do not announce that they are audiotaping. Oregon thus tramples on people’s ability to record and report on a large swath of public and newsworthy events. And because the law bans the taping of 74 PROJECT VERITAS V. SCHMIDT conversations where there is no reasonable expectation of privacy, Oregon’s statute is not narrowly tailored to further the state’s interest in conversational privacy. In any event, Oregon’s law should be subject to strict scrutiny, not intermediate scrutiny, because the statute is not content-neutral. The statute has a law-enforcement exception that allows citizens to legally record law enforcement officials—but no one else—without announcing that they are recording them. Oregon has essentially carved out only law enforcement matters from its ban on unannounced recording. Because this is a content- based restriction, strict scrutiny applies—and Oregon’s law must fall to the wayside. * * * * Under Oregon’s recording statute, a person may not record a conversation unless “all participants in the conversation are specifically informed” that they are being recorded. Or. Rev. Stat. § 165.540(1)(c). “Conversations” include any “transmission between two or more persons of an oral communication which is not a telecommunication or a radio communication, and includes a communication occurring through a video conferencing program.” Id. § 165.535(1). Because Oregon’s law even bans recording loud conversations in public—where there is no expectation of privacy—it is much broader than other states’ recording laws. Project Veritas v. Schmidt, 72 F.4th 1043, 1050 n.1 (9th Cir. 2023), vacated 95 F.4th 1152 (9th Cir. 2024). As one Oregon state court recognized, the law was intended to “protect[]participants in conversations from being recorded without their knowledge.” State v. Neff, 265 P.3d 62, 66 (Or. Ct. App. 2011). PROJECT VERITAS V. SCHMIDT 75 But despite its breadth, the law contains an exception for taping law enforcement. It permits recording “a conversation in which a law enforcement officer is a participant, if (A) The recording is made while the officer is performing official duties; (B) The recording is made openly and in plain view of the participants in the conversation; (C) The conversation being recorded is audible to the person by normal unaided hearing; and (D) The person is in a place where the person lawfully may be.” Or. Rev. Stat. § 165.540(5)(b). 1 * * * * I. Even under intermediate scrutiny, Oregon’s law cannot survive because it is not narrowly tailored to further conversational privacy. The First Amendment prohibits laws “abridging the freedom of speech.” U.S. Constit. amend I. That right includes the process of writing, drawing, and (in the modern era) creating a video or audio recording. We have recognized that the act of creating an expressive work— whether writing, drawing, or recording—is itself a form of protected speech. Anderson v. City of Hermosa Beach, 621 F.3d 1051, 1061-62 (9th Cir. 2010). And for good reason: If the government can regulate or ban the process of creating speech, then it wields the power to regulate or ban speech itself. Id. 1 Oregon also exempts recording “a conversation during a felony that endangers human life,” even if the recording begins before the felony. Id. § 165.540(5)(a); State v. Copeland, 522 P.3d 909, 912-13 (Or. Ct. App. 2022). This exception allows police officers who wear wires to record “situation[s] [that] involve[] felon[ies] where drugs are involved or human life is endangered.” H.B. 2252, 65th Assemb., Reg. Sess. 1 (Or. 1989). 76 PROJECT VERITAS V. SCHMIDT As the majority recognizes, we held that this right to record extends to surreptitious and non-consensual recordings in Wasden, 878 F.3d at 1203 (holding unconstitutional a state law that banned people “from entering a private agricultural production facility and, without express consent from the facility owner, making audio or video recordings of the ‘conduct of an agricultural production facility’s operations’”) (citing Idaho Code § 18- 7042(1)(d)). Despite Wasden’s holding, the majority upholds Oregon’s law, ruling that only intermediate scrutiny applies because the law is content-neutral (more on that later—it is not). But even assuming intermediate scrutiny applies, Oregon’s law does not pass constitutional muster because it is not narrowly tailored to further the state’s interest in conversational privacy. The statute does not ban just secret recordings in which someone believes that he or she is having a private conversation with someone—only to discover later that someone has been taping it. It also bans taping most conversations—even loud ones that happen in public or when there is no reasonable expectation of privacy—if a citizen has not alerted the speaker that he is being taped. Oregon’s law thus not only bans undercover journalism, it also might prohibit much of our modern-day news reporting. If, say, an innocent passerby records a local mayor loudly uttering a racial slur at someone on a public street, that person has violated Oregon’s law unless she announces to the mayor beforehand that she is recording it. So, too, for a patron who tapes an elected official launching a tirade against a Starbucks employee for failing to use organic, fair-trade coffee beans in his latte. These recordings are the types of newsworthy events that we would expect to PROJECT VERITAS V. SCHMIDT 77 see on the evening news or trending on social media. Yet Oregon, under the guise of advancing conversational privacy, criminalizes citizens for taping these events. But someone yelling on a public street or loudly causing a scene at a coffee shop has no reasonable expectation of privacy. Simply put, Oregon’s law is egregiously overbroad and is not narrowly tailored to protect conversational privacy, as it sweeps in all sorts of public and newsworthy events in which no privacy interest is at stake. See Berger v. City of Seattle, 569 F.3d 1029, 1059 (9th Cir. 2009) (explaining that under intermediate scrutiny “any regulation of speech” must be “carefully calibrated to solve those problems”). Indeed, Oregon’s law is so broad that it bans audiotaping even if the speaker notices that someone has a recording device in her hand. State v. Haase, 895 P.2d 813, 815 (Or. Ct. App. 1995). 2 The speaker would understand that there is no reasonable expectation of privacy when the other person has a tape-recorder. Yet Oregon bans such recordings unless the other person affirmatively announces that she is taping the conversation, despite the lack of any privacy interest in that situation. The lack of narrow tailoring becomes even more apparent when we see Oregon’s different and more permissive treatment of taping phone calls. For telephone calls, Oregon has adopted a single-party consent rule. It is lawful to tape a phone conversation so long as one 2 The majority states that we misread Haase, noting that the state appellate court had reversed an order excluding an audio-recording made by a police officer’s wearable microphone. Maj. Op. 7, n.2. But the court did so because the officer had announced before taping, “I need to tell you before you start that this conversation is being monitored by camera and by audio means.” Haase, 895 P.2d at 816. 78 PROJECT VERITAS V. SCHMIDT participant in the call “consents” to taping it (i.e., one person decides to tape the phone call). Or. Rev. Stat. § 165.540(1)(a). But Oregon’s single-party consent rule violates the principle of conversational privacy that the state invokes to justify the law here, as the other person on the telephone call never consented to being taped. If anything, people have a stronger privacy interest in a private phone call than a loud conversation in the public or at a store. ^[3] Oregon’s different and illogical treatment between telephone and in-person conversations highlights how Oregon’s laws are badly misaligned with the state’s conversational privacy interest—and how Section 165.540(1)(c) is not narrowly tailored to that interest. It is no answer to say that a journalist or citizen can avoid criminal liability by announcing that she is recording the conversation or encounter. That government-mandated intrusion into a conversation will almost certainly distort its candid and authentic nature. We recognize this reality even in our profession: One of our bedrock principles—the attorney-client privilege—turns on the belief that people will not speak candidly if they believe that someone else may hear that conversation. Mohawk Indus., Inc. v. Carpenter, 558 U.S. 100, 108 (2009) (“assuring confidentiality . . . encourages clients to make ‘full and frank’ disclosures to their attorneys, who are then better able to provide candid 3 Oregon presumably justifies its single-party consent rule for telephone calls on the basis that there is a difference between a recording made by a participant in the call and a third-party (who is not involved in that conversation). Maybe that is a reasonable distinction based on the state’s weighing of privacy interests and news value of disclosing telephone conversations. But Oregon oddly does not apply that same distinction to in-person conversations, underscoring that the law is not narrowly tailored. PROJECT VERITAS V. SCHMIDT 79 advice. . . .”). So too here. The government will essentially alter the content of a conversation if it requires someone recording it to make an announcement that he or she is taping the speakers. A recorded conversation after an announcement may be as candid as a scripted reality television show. Nor is it sufficient to say, as the majority does, that a citizen has “alternative channels to engage in journalistic speech activities” by using “traditional tools of investigative reporting” such as “talking with sources” and “reviewing records.” Maj. Op. 46. Perhaps journalists can still manage to report the news through other means, but Oregon is still confiscating an arrow out of their quiver. The government should not meddle in the methods of journalists and dictate how they can do their job. Talking with self-interested sources or reviewing records cannot substitute for rare moments of candor that may be captured and revealed on tape. Further, someone’s voice—the sound, tone, and emphasis—can convey more meaning than mere written words on a piece of paper. For example, the contempt (or jocular nature) in someone’s tone can alter the meaning of a statement. That is why recordings represent a “significant medium for the communication of ideas.” See Wasden, 878 F.3d at 1203 (quoting Joseph Burstyn, Inc. v. Wilson, 343 U.S. 495, 501 (1952)). An audiotape recording can also carry more weight than just words on a piece of paper, as a subject will find it harder to claim that he never said something or that he was misquoted (or conversely the audiotape may undercut the journalist’s characterization of a conversation). Faced with the prospect that Oregon’s law would impede a large chunk of everyday news reporting, the majority accepts the state’s unfounded assertion that its statute does 80 PROJECT VERITAS V. SCHMIDT not apply to “open recording” scenarios, citing to Section 165.540(6)(a)’s exceptions. But those exceptions do not broadly apply to any “open recordings.” Rather, the exceptions to the ban are specifically limited to narrow factual scenarios: (a) “[p]ublic or semipublic meetings such as hearings before governmental or quasi-governmental bodies, trials, press conferences, public speeches, rallies and sporting or other events,” (b) “[r]egularly scheduled classes or similar educational activities in public or private institutions,” and (c) “[p]rivate meetings or conferences if all others involved knew or reasonably should have known that the recording was being made.” Or. Rev. St. § 165.540(6)(a)(A)-(C). Oregon courts have interpreted those statutory terms under their ordinary and specific meaning, holding that the word “meetings” does not include “mere encounters.” State v. Bischel, 790 P.2d 1142, 1144 n.3 (Or. Ct. App. 1990) (“mere encounters are not ‘meetings’ within the meaning of [the statute]. Defendant’s argument that her taping of the confrontation with police was exempt either as a ‘public or semipublic meeting,’ or as a ‘private meeting or conference,’ . . . is without merit.”). Therefore, these “open recording” exceptions would not apply to any of the examples mentioned earlier because all of them are “mere encounters,” not “meetings” that are exempt from the ban. Oregon citizens thus could not, without an announcement, tape an elected official who bullies a barista at a Starbucks, a mayor who screams racial epithets on a public street, or a city council member who makes a damaging admission to a constituent at a mall even after seeing that the person has a tape-recorder in her hand. Audiotapes capturing these newsworthy events would never see the light of day under the majority opinion. PROJECT VERITAS V. SCHMIDT 81 II. The recording statute is content-based because it carves out an exception for law enforcement matters. A. The law-enforcement exceptions are not content-neutral and fail under strict scrutiny. As noted earlier, Oregon’s law cannot survive intermediate scrutiny, even assuming that is the proper tier of scrutiny. But we should be applying strict scrutiny here because the law is content-based. Barr v. Am. Assoc. of Pol. Consultants, Inc., 591 U.S. 610, 618 (2020). The First Amendment prohibits the government from “restrict[ing] expression because of its message, its ideas, its subject matter, or its content.” Police Dept. of City of Chicago v. Mosley, 408 U.S. 92, 95 (1972). Any law that “target[s] speech” based on its message, ideas, subject matter, or content is “presumptively unconstitutional.” Reed v. Town of Gilbert, 576 U.S. 155, 163 (2015). A law is content- based, and thus presumptively unconstitutional, if it “applies to particular speech because of the topic discussed or the idea or message expressed.” Id. To determine whether a law is content-based, we look to whether the law “single[s] out any topic or subject matter for differential treatment.” City of Austin v. Reagan Nat’l Advert. of Austin, LLC, 596 U.S. 61, 71 (2022). Oregon’s law-enforcement exceptions are necessarily content-based because they “single out a[] topic or subject matter”—law enforcement—“for differential treatment.” Id. By its plain language, the statute allows citizens to record— without making an announcement—a police officer performing his or her official duties. Or. Rev. Stat. § 165.540(5)(b). No similar exception exists for any other profession or field. It is unlawful for citizens to tape 82 PROJECT VERITAS V. SCHMIDT (without an announcement) elected officials, schoolteachers, public housing agency administrators, environmental services employees, tax collectors, department of water and power representatives, child welfare investigators, and anyone else. But it is lawful to record law enforcement officials—and only law enforcement officials—without any announcement. 4 The majority argues that the statute is not content-based because “the requirement in § 165.540(5)(b) that a law enforcement officer be involved in the conversation does not regulate a ‘topic’ because the statute is unconcerned with the content of the conversation in which an officer participates.” Maj. Op. 33, n.13. That assertion defies common sense and the statutory language. That is like saying that a law that bans Hollywood from featuring law enforcement officers— or anti-police brutality activists—in its movies is not content-based because it does not regulate a specific topic or conversation in the movie. The whole point of the statutory exception is to carve out law enforcement as a subject matter by allowing citizens to tape officers during their official duties without making an announcement. Notably, the Oregon statute does not create a blanket exception to tape anything that a law enforcement officer may say at any time; it only applies to taping a law enforcement officer “performing official duties.” Or. Rev. 4 Oregon’s law thus may not just be content-based but also viewpoint- based, because critics of the police are given a speech tool that the statute denies to critics of other officials. See Rosenberger v. Rector and Visitors of Univ. of Virginia, 515 U.S. 819, 829 (1995) (“When the government targets not subject matter, but particular views taken by speakers on a subject, the violation of the First Amendment is all the more blatant. Viewpoint discrimination is thus an egregious form of content discrimination”) (internal citation omitted). PROJECT VERITAS V. SCHMIDT 83 Stat. § 165.540(5)(b)(A). And if a police officer is “performing official duties,” what else could he or she be talking about other than law enforcement matters? Law enforcement matters are thus the very “topic” or “content” that the law targets. The majority’s contention that the law “draws a line based on the circumstances in which a recording is made, not on the content of the conversation recorded” (Maj. Op. 34) is belied by the statute’s exception for only “official duties.” Consider this example: It would be illegal to audio-record a police officer at a coffee shop countertop loudly talking about Portlandia (because the officer is not “performing official duties”), but someone can lawfully record that same police officer if he interrogates a suspect at that same coffee shop (because he is performing “official duties”). The only difference between the two scenarios is that the officer is talking about a TV show in the first example but about law enforcement matters in the second example. In sum, recording a police officer performing his or her official duties will necessarily be about that officer’s law enforcement duties—and thus law enforcement will be the “topic” or “content” that Oregon’s law targets. That is a content-based restriction. The content-based nature of Oregon’s law becomes even clearer if we consider other analogous scenarios. Assume Oregon enacts a similar carve-out for labor union officials— it is lawful to record a labor union official while he or she is “performing official duties” (but nobody else). It would be obvious that the law is content-based, i.e., topics pertaining to labor unions are treated differently from other topics. Or assume that Oregon bans non-consensual audiotaping except that citizens can lawfully record officials at the University of Oregon’s Division of Equity and Inclusion who are “performing their official duties.” Such a distinction would 84 PROJECT VERITAS V. SCHMIDT be content-based, as it treats some topics differently from others. We would not say such a law is not content-based because “the statute is unconcerned with the content of the conversation in which [a labor union official or DEI personnel] participates.” Cf. Maj. Op. at 33, n. 13. The majority also argues that the Oregon law does not discriminate based on content, relying on City of Austin. Maj. Op. at 33. But that case is easily distinguishable. The Supreme Court held only that, when a “substantive message” is “irrelevant to the application of” a law, the need to consider the message itself does not necessarily make the law content-based. City of Austin v. Reagan National Advertising of Austin, LLC, 596 U.S. 61, 71 (2022). And that makes sense: In City of Austin, a municipality had set different rules for on-site and off-site business advertisements. Id. A business argued that, because an enforcer would need to read a sign to see whether it was advertising an on-site or off-site business, the law was content-based. Id. But the Supreme Court rejected this argument because the content of the sign “matter[ed] only to the extent” it matches the location of the sign. Id. In other words, the City of Austin regulations were not targeting content—they were regulating the placement of content. Id. And the Supreme Court has long held that time, place, and manner restrictions like those are content-neutral. Oregon argues that, just as in City of Austin, its law turns on a content-neutral distinction—the activity of recording. But what “activity” does the law target? It is not the manner of recording—the law applies equally to someone who wears a wire, uses a tape recorder, live-streams on Instagram, or takes out a tripod and professional camera. Cf. Bartnicki v. Vopper, 532 U.S. 514, 526 (2001). Nor is it the way someone acts while recording a conversation—the law PROJECT VERITAS V. SCHMIDT 85 applies the same to someone who fails to announce that he or she is recording a loud argument while standing on a sidewalk as it does to someone who, while out to eat, secretly records a nearby dinner table’s quiet conversation. Because Oregon’s law imposes content-based restrictions on speech, it is unconstitutional unless Oregon shows that it can survive strict scrutiny. Reed, 576 U.S. at 166. To do so, Oregon must show that the law is narrowly tailored—meaning, necessary—to serve a compelling interest. R.A.V. v. St. Paul, 505 U.S. 377, 395–96 (1992). As noted earlier, Oregon’s law is not narrowly tailored. The law covers conversations in which there is no privacy interest or in which the privacy interest is outweighed by the newsworthiness of the conversation. The law thus applies to more speech than is necessary to serve its interest, and it cannot survive strict scrutiny. B. The law-enforcement exceptions are not severable. The majority concludes that, even if the law-enforcement exceptions are content-based, they are severable and thus do not require that we find the law unconstitutional. Maj. Op. at 47-50. We look at state law to determine severability. Leavitt v. Jane L., 518 U.S. 137, 139 (1996). And under Oregon law, we look at whether “parts of the statute are so interconnected that it appears likely that the remaining parts would not have been enacted without the unconstitutional part, or . . . [if] the remaining parts are incomplete and cannot be executed in accordance with legislative intent.” Outdoor Media Dimensions, Inc. v. Dep’t of Transp., 132 P.3d 5, 18 (Or. 2006). In our case, we must ask whether Oregon would have enacted its ban on non-consensual recording if it could not retain its exceptions for recording 86 PROJECT VERITAS V. SCHMIDT law enforcement officers and acts of felony that endanger human life. It is impossible to sever those two content-based carve- outs. We have held that citizens have a right to “record law enforcement officers engaged in the exercise of their official duties in public places.” Askins v. U.S. Dept. of Homeland Sec., 899 F.3d 1035, 1044 (9th Cir. 2018). We held that this right was “include[d]” within the broader First Amendment “right to photograph and record matters of public interest.” Id. Severing this law enforcement exception would thus merely substitute one constitutional deficiency for another. See id. There is no basis to conclude that the Oregon Legislature would have enacted the much broader—and constitutionally deficient—statute that would result from severing this content-based exception. The majority notes that Oregon enacted this ban on non- consensual recording in 1959—and only much later did it enact the two exceptions. Maj. Op. at 48. The majority thus reasons that Oregon would enact this statute without the exceptions (because it did so). But in 1959 we had not issued a decision stating that there is a right to record law enforcement officers performing their official duties. Oregon later enacted the law enforcement exception because it recognized that its broad law raised constitutional issues under judicial precedent (but in doing so it created a different problem of imposing a content-based restriction). Likewise, severing the felony exception could not save the law, as the law would still unconstitutionally prohibit “record[ing] matters of public interest.” Id. Finally, the majority argues that Oregon would be placed in an “insoluble dilemma” under this dissent’s reasoning because its law would be struck down as unconstitutional PROJECT VERITAS V. SCHMIDT 87 either on content-based or broader First Amendment grounds. Maj. Op. at 49. Not so. Oregon could have enacted a narrowly tailored law that limited its audio- recording ban to conversations where there is a reasonable expectation of privacy. But it declined to do so. In any event, under Oregon’s limited severability analysis, we only look at whether the state would have enacted this statute without the unconstitutional content-based carve-out. It does not appear Oregon would have done so because it would have then faced potential First Amendment problems without such a carve-out. In short, this statute cannot be saved by severability. CONCLUSION Oregon’s law violates the First Amendment by barring unannounced taping of conversations that occur loudly in public or in which no privacy interest is at stake. I fear that it will hamper basic reporting of public and newsworthy events. I thus respectfully dissent.