IKUTA, Circuit Judge:

Oregon law generally prohibits unannounced recordings of conversations, subject to several exceptions. We conclude that Oregon’s law is a content-based restriction that violates the First Amendment right to free speech and is therefore invalid on its face. I A Section 165.540(1)(c) of the Oregon Revised Statutes provides: “[A] person may not . . . [o]btain or attempt to obtain the whole or any part of a conversation by means of any device . . . if not all participants in the conversation are specifically informed that their conversation is being

6 PROJECT VERITAS V. SCHMIDT

obtained.” Or. Rev. Stat. § 165.540(1)(c). 1 The statute defines “[c]onversation” 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. Stat. § 165.535(1). Because this section explicitly applies to the recording of a video conference and bars individuals from obtaining a conversation “by means of any device,” it applies to both audio and video recordings of a conversation. Indeed, the Oregon courts have interpreted the statute as applicable to video recordings of conversations and other conduct. 2 See State v. Copeland, 522 P.3d 909, 911–12 (Or. Ct. App. 2022) (applying section 165.540(1)(c) to “the video and audio recording of [a] shooting taken by the victim on his body camera”). 3 Wasden again guides our analysis. After holding that the creation of audio and video recordings was speech entitled to full First Amendment protection, Wasden held that the Idaho statute at issue in that case, which required the facility owner’s consent to make unannounced recordings of “the conduct of an agricultural production facility’s operations,” was “an ‘obvious’ example of a content-based regulation of speech because it ‘defin[es] regulated speech by particular subject matter.’” 878 F.3d at 1204 (citing Reed, 576 U.S. at 163). We gave two reasons for this conclusion. First, the statute drew “a distinction ‘on its face’ regarding the message the speaker conveys.” Id. (citing Reed, 576 U.S. at 165). Specifically, it “would permit filming a vineyard’s art collection but not the winemaking operation.” Id. “Likewise, a videographer could record an after-hours birthday party among co-workers, a farmer’s antique car collection, or a historic maple tree but not the animal abuse, feedlot operation, or slaughterhouse conditions.” Id. Second, we reasoned that “only by viewing the recording can the [state] authorities make a determination about criminal liability” because the application of the exception “explicitly pivots on the content of the recording.” Id. Our second rationale (that a law regulating the act of making specified recordings is content based if state authorities cannot apply the law without viewing or listening to the particular recording at issue) requires some further examination. After we decided Wasden, the Supreme Court rejected a per se rule “that a regulation cannot be content neutral if it requires reading the [speech] at issue.” City of Austin, 142 S. Ct. at 1471. Instead, City of Austin held that location-based rules, such as a rule differentiating between signs on a premise that advertise an on-site business from

PROJECT VERITAS V. SCHMIDT 19

signs that advertise some off-site matter, are not content based, even though city authorities had to review the sign’s message to apply the rule. Id. at 1472. When a rule is merely a “location-based and content-agnostic on-/off-premises distinction,” it does not “singl[e] out specific subject matter for differential treatment.” Id. at 1475 (citation omitted). Instead, the sign’s message merely “informs the sign’s relative location.” Id. at 1473. But as the Court clarified, this exception for location-based rules does not affect the Court’s longstanding holding that “regulations that discriminate based on the topic discussed or the idea or message expressed . . . are content based.” Id. at 1474 (citation and quotation marks omitted). Wasden did not address a location-based rule akin to an “on-/off-premises distinction,” but considered a rule that singled out “specific subject matter for differential treatment” and discriminated based on “the topic discussed or the idea or message expressed.” Id. at 1474–75. As a result, City of Austin’s analysis does not conflict with our holding in Wasden, which remains binding. See Miller v. Gammie, 335 F.3d 889, 893 (9th Cir. 2003) (holding that a three-judge panel is bound by precedent unless it “is clearly irreconcilable with the reasoning or theory of intervening higher authority”). Therefore, we continue to consider whether a law “pivots on the content of the recording,” Wasden, 878 F.3d at 1204, in determining whether the law discriminates on the basis of “the topic discussed or the idea or message expressed” and is, therefore, content based, City of Austin, 142 S. Ct. at 1474 (citing Reed, 576 U.S. at 171). Applying Wasden here, section 165.540 is a content- based restriction on speech. On its face, section 165.540(1)(c) and its exceptions draw a distinction between topics. The speech regulated by section 165.540(1)(c) is the

20 PROJECT VERITAS V. SCHMIDT

act of making a recording, which means that the activity captured by a recording constitutes the content or subject matter of that speech. Because the rules imposed by section 165.540 vary depending on the activity being recorded, the statute clearly draws content-based distinctions under Wasden. The law’s applicability plainly “pivots on the content of the recording”—namely, what the recording captures. Wasden, 878 F.3d at 1204. For example, the law applies no restrictions to recording law enforcement officials engaged in their official duties, see Or. Rev. Stat. § 165.540(5)(b), but prohibits recording other government officials performing official duties unless they are informed that their conversation is being recorded. Similarly, the statute distinguishes between recording felonies endangering human lives, id. § 165.540(5)(a), and recording similar conduct during the commission of a misdemeanor. These distinctions are “obvious” examples of a content- based regulation of speech because they “define regulated speech by particular subject matter.” Wasden, 878 F.3d at 1204 (cleaned up) (citation omitted). In addition, state “authorities [can] make a determination about criminal liability” under the law “only by viewing the recording.” Id. This serves as further evidence that the applicability of section 165.540(1)(c) pivots on the content of the recording, thereby demonstrating that the law is content based.

Oregon argues that section 165.540(1)(c)’s general prohibition on the act of making unannounced recordings is a content-neutral speech regulation for two reasons. Neither is persuasive. Oregon first argues that the statute is content neutral because the statute’s exceptions are not based on the words

PROJECT VERITAS V. SCHMIDT 21

spoken and recorded, and therefore state authorities do not have to listen to and analyze the recording to determine whether an exception applies. We disagree. The statute at issue in Wasden did not distinguish based on the words spoken in a recording, but we nevertheless held that it was content based because it discriminated on the basis of subject matter to be recorded. 878 F.3d at 1204. For the same reason, it is the statute’s differential treatment of recordings based on their subject matter (e.g., whether the speaker’s recording obtains the conversation of Oregon police officers or Oregon executive officers) that makes the statute content based, not the words exchanged in the conversation. Second, Oregon argues that we can consider section 165.540(1)(c) as a stand-alone provision, and ignore the exceptions to the general prohibition. But this approach is foreclosed by binding precedent. To start, it is well- established that when a court evaluates the constitutionality of a general prohibition, it must consider any exceptions to the general rule. “[A] rule [is] content-based when it establishes a general ban on speech, but maintains exceptions for speech on certain subjects.” Glendale Assocs., Ltd. v. NLRB, 347 F.3d 1145, 1155 (9th Cir. 2003). Stated differently, where exceptions to a restriction of protected speech “are based on content, the restriction itself is based on content.” Nat’l Ad. Co. v. City of Orange, 861 F.2d 246, 249 (9th Cir. 1988) (citation omitted); see also Barr v. Am. Ass’n of Pol. Consultants, Inc., 140 S. Ct. 2335, 2347 (2020) (plurality opinion) (holding that a prohibition of robocalls was content based due to its exception for robocalls collecting government held debts); Foti, 146 F.3d at 636 (holding that an ordinance’s general ban of “all signs on all public property” was content based due to its

22 PROJECT VERITAS V. SCHMIDT

“exemptions for open house signs and safety, traffic, and public informational signs”).11 Moreover, any exception to a general restriction on protected speech—even if the exception applies to speech that our case law has recognized as receiving First Amendment protection, like recording police officers performing official duties in public, see Fordyce, 55 F.3d at 439; Askins v. U.S. Dep’t of Homeland Sec., 899 F.3d 1035, 1044 (9th Cir. 2018)—necessarily renders the restriction content based. The Supreme Court analyzed a similar situation in Reed, where the challenged state law generally restricted the display of outdoor signs without a permit, but exempted signs that had ideological and political messages, which implicate speech that case law has recognized as receiving First Amendment protection. 576 U.S. at 164–65. Despite these exceptions, the Court held that the law as a whole was content based and subject to strict scrutiny, “regardless of the government’s benign motive, content- neutral justification, or lack of ‘animus toward the ideas

11 The district court concluded that the law enforcement exception did not render section 165.540(1)(c) content based because recordings of “conversations where a law enforcement officer is a speaker” is “government speech,” which “is generally not subject to First Amendment challenges.” Oregon does not rely on this argument, and we conclude the government speech doctrine is not applicable here. Although the Supreme Court has held that a government entity’s expression of its own views does not violate the speech rights of individuals who disagree, see Pleasant Grove City v. Summum, 555 U.S. 460, 467 (2009), this case does not involve a suit against the government for expressing its views. Rather, it involves a statute that impinges on a private individual’s speech by restricting the ability to record public officials. The individual engaging in the speech being regulated is the private party that makes the recording—not the government. Therefore, the government speech doctrine is inapposite.

PROJECT VERITAS V. SCHMIDT 23

contained’ in the regulated speech.” Id. at 165 (citation omitted). Therefore, under this precedent, we must analyze both the general prohibition and the exceptions as one regulatory regime. Doing so makes clear that section 165.540 is a content-based regulation of speech.12 C Because we conclude that section 165.540(1)(c) and its exceptions constitute a content-based speech restriction, we can uphold the statute only if it survives strict scrutiny. See Wasden, 878 F.3d at 1204. Strict scrutiny requires the government to show that the speech restriction is “narrowly tailored to address the State’s compelling governmental interests.” Victory Processing, LLC v. Fox, 937 F.3d 1218, 1229 (9th Cir. 2019). Under strict scrutiny, the challenged law must be constitutional with respect to “each activity within the proscription’s scope.” Berger v. City of Seattle, 569 F.3d 1029, 1053 (9th Cir. 2009) (citing Frisby v. Schultz, 487 U.S. 474, 485 (1988)). It does not matter that a narrower restriction on speech activities could have been justified by the government’s interest. See Watchtower Bible & Tract Soc’y of N.Y., Inc. v. Village of Stratton, 536 U.S. 150, 168 (2002). For instance, a law that generally prohibits canvassers from engaging in door-to-door advocacy without a permit is facially unconstitutional. Id. Although the government’s “interest in preventing fraud could adequately support the ordinance insofar as it applies to commercial transactions and the solicitation of funds,” the interest in fraud prevention does not justify the ordinance insofar as it

12 The dissent concedes that the statutory exceptions to the general ban on unannounced recordings render section 165.540 content based. Dissent at 50.

24 PROJECT VERITAS V. SCHMIDT

applies “to [Jehovah Witnesses who offer religious literature], to political campaigns, or to enlisting support for unpopular causes.” Id. In Wasden, we applied strict scrutiny to the content- based Idaho statute. 878 F.3d at 1204. We assumed that Idaho’s asserted interest in protecting both property and privacy interests in an agricultural production facility was a compelling government interest, see id., but concluded that Idaho had not satisfied the narrow tailoring requirement because, among other reasons, there were “various other laws at Idaho’s disposal that would allow it to achieve its stated interests while burdening little or no speech,” id. at 1205 (cleaned up) (citation and quotation marks omitted). “For example, agricultural production facility owners can vindicate their rights through tort laws against theft of trade secrets and invasion of privacy.” Id. And, as another example, “[t]o the extent the legislators expressed concern that fabricated recordings of animal abuse would invade privacy rights, the victims can turn to defamation actions for recourse.” Id. Further, we explained, “‘the remedy for speech that is false is speech that is true’—and not, as Idaho would like, the suppression of that speech.” Id. (cleaned up) (citation omitted). Therefore, we struck down Idaho’s ban on creating audio and visual recordings as failing to survive First Amendment scrutiny. Id. Applying strict scrutiny to section 165.540(1)(c) in light of these precedents, we must consider whether that section is constitutional with respect to “each activity within the proscription’s scope,” Berger, 569 F.3d at 1053, which necessarily includes its regulation of protected speech in places open to the public, see supra pp. 14 n.9, 23–24.

PROJECT VERITAS V. SCHMIDT 25 We first consider the nature of Oregon’s interest here. At the outset, Oregon does not assert it has a compelling interest, but argues only that it has a significant governmental interest in protecting individuals’ conversational privacy. In analyzing this interest, we are bound by Wasden’s conclusion that “[t]he act of recording is itself an inherently expressive activity” that merits First Amendment protection. 878 F.3d at 1203. Therefore, prohibiting a speaker’s creation of unannounced recordings in public places to protect the privacy of people engaged in conversation in those places is the equivalent of prohibiting protesters’ or buskers’ speech in public places for the same purpose. See Berger, 569 F.3d at 1054. Thus, we must analyze Oregon’s interest in conversational privacy as protecting people’s conversational privacy from the speech of other individuals, even in places open to the public. In general, the government does not have a compelling interest in protecting individual privacy against unwanted communications (including the “speech” comprised of recording others) in areas open to the public unless the audience’s “substantial privacy interests are being invaded in an essentially intolerable manner.” Cohen v. California, 403 U.S. ^[15], 21 (1971); see also Hill v. Colorado, 530 U.S. 703, 717 (2000) (recognizing that the government’s interest in protecting privacy “varies widely in different settings”). Courts have recognized such a compelling interest only when patients seeking medical care are bombarded by “the cacophony of political protests” and individuals at their homes are confronted with unwanted speech. Hill, 530 U.S. at 716. The government’s interest in protecting the public’s privacy from unwanted speech (including recordings of people’s conversations) “is far less important” for

26 PROJECT VERITAS V. SCHMIDT

individuals engaging in recreational, social, or commercial activities in places open to the public, such as “strolling through Central Park,” id., or “waiting in line or having lunch outdoors in a public park,” Berger, 569 F.3d at 1054. Indeed, we have held that the government does not even have a “significant interest in protecting [individuals] from unpopular speech” where those who constitute the intended audience are commercial patrons of “a place of public entertainment.” Kuba v. 1-A Agric. Ass’n, 387 F.3d 850, 861 n.10 (9th Cir. 2004). Applying this framework here, Oregon does not have a compelling interest in protecting individuals’ conversational privacy from other individuals’ protected speech in places open to the public, even if that protected speech consists of creating audio or visual recordings of other people.

Nor is Oregon’s rule narrowly tailored to be “the least restrictive or least intrusive means of” achieving the government’s interest in conversational privacy, as required to pass strict scrutiny review. Ward v. Rock Against Racism, 491 U.S. 781, 798–99 & n.6 (1989). Under strict scrutiny, a speech restriction must “target[] and eliminate[] no more than the exact source of the ‘evil’ it seeks to remedy.” Frisby, 487 U.S. at 485 (citation omitted). A law is not narrowly tailored if it restricts “speech that do[es] not cause the types of problems that motivated the [law].” Comite de Journaleros de Redondo Beach v. City of Redondo Beach, 657 F.3d 936, 948 (9th Cir. 2011) (en banc). In addition, a law is not narrowly tailored if it is over-inclusive because it suppresses more speech than is necessary to further Oregon’s goal of protecting people’s conversational privacy. See Wasden, 878 F.3d at 1205.

PROJECT VERITAS V. SCHMIDT 27

Applying this test, we conclude that section 165.540 burdens more protected speech than is necessary to achieve its stated interest. See id. The law regulates protected speech to avoid impinging on people’s conversational privacy. But in public places, speech does not intrude on privacy unless it intrudes in “an essentially intolerable manner.” See Berger, 569 F.3d at 1056 (holding that a statute prohibiting “passive and unthreatening acts” such as offering a handbill or displaying a sign, even if the communications were unwanted, was not narrowly tailored under intermediate scrutiny). As the Supreme Court has explained, “it is difficult, indeed, to justify a prohibition on all uninvited approaches . . . regardless of how peaceful the contact may be, without burdening more speech than necessary to prevent intimidation.” Madsen v. Women’s Health Center, Inc., 512 U.S. 753, 774 (1994). Section 165.540(1)(c) does not distinguish between “passive and unthreatening” acts and intolerable intrusions. Under our case law, that does not constitute narrow tailoring. Moreover, where speech occurs in places open to the public, the privacy interest of other individuals in those public areas is implicated only if and where the speech is unwanted. See Hill, 530 U.S. at 716; Berger, 569 F.3d at 1056. Yet section 165.540(1)(c) does not distinguish between wanted and unwanted speech (including wanted or unwanted recordings).13 For example, protesters demonstrating in favor of their political views may have no objection to unannounced recordings of conversations, which would provide more publicity about their goals and

13 For its part, the dissent apparently assumes without explanation that all unannounced recordings are unwanted speech and all announced recordings are welcomed speech. Dissent at 53–61.

28 PROJECT VERITAS V. SCHMIDT

beliefs. While some people may desire privacy for a conversation held in places open to the public, such instances cannot justify Oregon’s wholesale restriction on protected speech (i.e., recordings) in public areas. See Rock Against Racism, 491 U.S. at 799 (stating that a speech restriction “may not regulate expression in such a manner that a substantial portion of the burden on speech does not serve to advance its goals”). The dissent argues that Berger and its progeny are inapplicable to section 165.540(1)(c) because “state action aimed at protecting people from unwanted commercial or political speech” is “qualitatively different” than state action protecting people “from speech-gathering activities like Project Veritas’s . . . because they appropriate the speech of others.” Dissent at 45. According to the dissent, the sort of speech that includes the “appropriation of another person’s speech” (i.e., recordings) is qualitatively more burdensome than other types of speech that might intrude on a person’s privacy. Dissent at 64. This position is foreclosed by Wasden, which did not accord any special attention to the privacy interests of people whose speech might be recorded. Rather, Wasden held that a state law prohibiting audio or video recordings of the conduct of an agricultural production facility’s operations, which necessarily would include conversations, directions, and other forms of oral communications, “suppresse[d] more speech than necessary to further Idaho’s stated goals of protecting property and privacy.” 878 F.3d at 1205. Wasden’s analysis of recordings under the same framework applicable to other sorts of protected speech is consistent with precedent: for example, under our case law, we analyze expressive conduct that merits First Amendment protection as symbolic speech in the same manner as we analyze oral

PROJECT VERITAS V. SCHMIDT 29

communications. See Swisher, 811 F.3d at 318 (“Content- based prohibitions of speech and symbolic speech are analyzed under the same framework.”).14 Finally, as in Wasden, the rule is not narrowly tailored because “there are various other laws at [Oregon’s] disposal that would allow it to achieve its stated interests while burdening little or no speech.” 878 F.3d at 1205 (citation and quotation marks omitted). Individuals whose conversation is captured in public by unannounced recordings “can vindicate their rights” through an invasion of privacy tort. See, e.g., Humphers v. First Interstate Bank of Oregon, 696 P.2d 527, 531–32 (Or. 1985) (en banc) (noting that Oregon has recognized the common law privacy torts of appropriation, offensive publication of private facts, and intrusion upon exclusion); State v. Lien, 441 P.3d 185, 193 (Or. 2019) (“Tortious invasion of privacy is one of the limited number of torts in Oregon in which a plaintiff may be awarded damages consisting solely of mental suffering caused by the violation.”); Anderson v. Fisher Broad. Cos., 712 P.2d 803, 814 (Or. 1986) (explaining instances where a television program airing photographs of an accident victim could give rise to a tortious invasion of privacy claim);

14 The dissent’s reliance on Hurley v. Irish-American Gay, Lesbian & Bisexual Group of Boston, 515 U.S. 557 (1995), for the argument that speech involving the creation of a recording that captures people’s speech “implicates the ‘principle of autonomy to control one’s own speech’” is misplaced. Dissent at 57–58. Hurley held that the First Amendment prohibits the state from forcing a speaker to incorporate a message that the speaker does not want to convey. See id. at 559, 581. To the extent Hurley has any bearing on this case, it supports our view that a law raises serious constitutional issues if it prohibits a speaker from conveying the message the speaker wants to convey—candid responses to issues of controversy—by making unannounced recordings.

30 PROJECT VERITAS V. SCHMIDT

McLain v. Boise Cascade Corp., 533 P.2d 343, 345–46 (Or. 1975) (holding that unannounced recordings of the plaintiff “engaged in various activities on his property outside his home” were not actionable as invasion of privacy torts because the recordings “were done in such an unobtrusive manner that plaintiff was not aware that he was being watched and filmed” and the plaintiff “could have been observed by . . . [a] passerby”). Or if the recording is fabricated, “the victims can turn to defamation actions for recourse.” Wasden, 878 F.3d at 1205; see also Neumann v. Liles, 369 P.3d 1117, 1120–21 (Or. 2016).15 We conclude that section 165.540(1)(c) regulates protected speech (unannounced audiovisual recording), and is content based because it distinguishes between particular topics by restricting some subject matters (e.g., a state executive officer’s official activities) and not others (e.g., a police officer’s official activities). As a content-based restriction, the rule fails strict scrutiny review because the law is not narrowly tailored to achieving a compelling governmental interest in protecting conversational privacy with respect to “each activity within the proscription’s scope,” Berger, 569 F.3d at 1053, and there are other ways Under Outdoor Media, we may consider whether severing the exceptions to section 165.540(1)(c) would “save” that section’s general prohibition, even though the exceptions are not themselves unconstitutional. Assuming that section 165.540(1)(c), considered by itself, is content neutral, it can be “saved” as constitutional if it qualifies as a valid time, place, or manner restriction. Such a restriction must (1) be content neutral, (2) survive intermediate scrutiny review, and (3) “leave open ample alternative channels for communication of the information.” Hoye v. City of Oakland, 653 F.3d 835, 844 (9th Cir. 2011) (citing Rock Against Racism, 491 U.S. at 791); see also Regan v. Time, Inc., 468 U.S. 641, 648 (1984). Assuming that section 165.540(1)(c) would be content neutral if it were a stand- alone provision and would survive intermediate scrutiny review, we conclude it does not satisfy the third requirement.

PROJECT VERITAS V. SCHMIDT 35

“[A] regulation that forecloses an entire medium of public expression across the landscape of a particular community or setting fails to leave open ample alternatives.” United Bhd. of Carpenters & Joiners of Am. Loc. 586 v. NLRB, 540 F.3d 957, 969 (9th Cir. 2008). Regulations may not hamper a speaker’s preferred mode of communication to such an extent that they compromise or stifle the speaker’s message. See McCullen v. Coakley, 573 U.S. 464, 487–90 (2014). Alternatives that are “less effective media for communicating the [speaker’s] message . . . . are far from satisfactory.” Linmark Assocs., Inc. v. Township of Willingboro, 431 U.S. 85, 93 (1977). “[F]ree speech protections extend to the right to choose a particular means or avenue of speech in lieu of other avenues.” United Bhd., 540 F.3d at 969 (cleaned up) (citation and quotation marks omitted). Thus, while the “[g]overnment may regulate the manner of speech in a content-neutral way,” the government “may not infringe on an individual’s right to select the means of speech.” Foti, 146 F.3d at 641–42. For example, in City of Ladue v. Gilleo, the Supreme Court held that an ordinance that prohibited displaying signs in front of one’s residence did not leave open ample alternative channels of communication. 512 U.S. 43, 56 (1994). In reaching that conclusion, the Supreme Court rejected the city’s argument that the law left open ample alternative channels of communication because “residents remain free to convey their desired messages by other means, such as hand-held signs, letters, handbills, flyers, telephone calls, newspaper advertisements, bumper stickers, speeches, and neighborhood or community meetings.” Id. (citation and quotation marks omitted). In doing so, the Supreme Court explained that “[d]isplaying a sign from one’s own residence often carries a message quite distinct

36 PROJECT VERITAS V. SCHMIDT

form placing the same sign someplace else, or conveying the same text or picture by other means.” Id. Indeed, it is “[p]recisely because of their location [that] such signs provide information about the identity of the speaker.” Id. (quotation marks omitted). To illustrate, the Supreme Court noted that “[a] sign advocating ‘Peace in the Gulf’ in the front lawn of a retired general or decorated war veteran may provoke a different reaction than the same sign in a 10-year- old child’s bedroom window or the same message on a bumper sticker of a passing automobile.” Id. Likewise, “[a]n espousal of socialism may carry different implications when displayed on the grounds of a stately mansion than when pasted on a factory wall or an ambulatory sandwich board.” Id. at 56–57. Moreover, the intention behind placing a sign at one’s residence may be “to reach neighbors, an audience that could not be reached nearly as well by other means.” Id. at 57 (emphasis omitted). In some instances, barring a means of speech effectively eliminates a message. For speakers “of modest means or limited mobility, a yard or window sign may have no practical substitute.” Id. And for others, “the added costs in money or time of taking out a newspaper advertisement, handing out leaflets on the street, or standing in front of one’s house with a handheld sign may make the difference between participating and not participating in some public debate.” Id. In light of this understanding of what case law requires for a speech restriction to leave open ample alternative channels for communication, it is clear that section 165.540(1)(c) does not meet the mark. It functions as “an absolute prohibition on a particular type of expression”—the creation of unannounced audiovisual recordings. United States v. Grace, 461 U.S. 171, 177 (1983). Though section 165.540(1)(c) allows individuals to record conversations

PROJECT VERITAS V. SCHMIDT 37

where participants are “specifically informed that their conversation is being obtained,” such notification would effectively destroy the intended content of the recording. The subject matter of unannounced recordings is the subjects’ candid responses to issues of controversy. Because the protected speech is the recording of subjects’ unfiltered responses, see Wasden, 878 F.3d at 1204, a rule that requires the person creating the recording to provide notice extinguishes that speech. In other words, creating announced recordings is not an adequate alternative channel of speech for creating unannounced recordings.19 Nor does after-the-fact reporting of an undercover interview or encounter provide an adequate alternative method of communication. Audiovisual recording is a unique medium of communication. It captures in real time both the sounds and sights of an event, making it more trustworthy and persuasive—and thus having vastly greater impact—than post-hoc written or oral accounts. See Fields v. City of Philadelphia, 862 F.3d 353, 359 (3d Cir. 2017) (noting that audiovisual recordings “corroborate[] or lay[] aside subjective impressions for objective facts”); Am. C.L. Union of Illinois v. Alvarez, 679 F.3d 583, 595, 607 (7th Cir. 2012) (stating that the “self-authenticating character” of audiovisual recordings “makes it highly unlikely that other methods could be considered reasonably adequate substitutes”). Indeed, the Supreme Court recognized the importance of audiovisual recording as corroborating or disproving testimony in Scott v. Harris. Even on summary

19 In fact, the dissent expressly acknowledges these attributes, which are unique to unannounced recordings. Dissent at 55. But by recognizing that unannounced recordings are unique, the dissent has necessarily conceded that other forms of media are inadequate substitutes.

38 PROJECT VERITAS V. SCHMIDT

judgment when “courts are required to view the facts and draw reasonable inferences ‘in the light most favorable to the party opposing the [summary judgment] motion,’” the court must rely on “the record of a videotape capturing the events in question,” when it “clearly contradicts the version of the story told by” the nonmoving party. 550 U.S. 372, 378 (2007) (citation omitted). Audiovisual recordings are also unique because they can readily be disseminated to a wider audience when incorporated into news programming. See Fields, 862 F.3d at 359 (“Recordings also facilitate discussion because of the ease in which they can be widely distributed via different forms of media.”); Am. C.L. Union of Illinois, 679 F.3d at 607 (noting that audiovisual recordings are “powerful methods of . . . disseminating news and information”). Accordingly, section 165.540(1)(c) does not leave open alternative channels to real-time, unannounced audiovisual recordings. And we therefore conclude that section 165.540(1)(c) (if read as a stand-alone provision, without exceptions) is not a valid time, place, or manner restriction. In opposing this analysis, and arguing that section 165.540(1)(c) leaves open ample alternative channels of communication, the dissent reframes the medium of public expression sought by Project Veritas at a high level of generality. According to the dissent, the relevant medium of communication is not the unannounced recordings that capture candid responses, but rather “investigative journalism” generally. Dissent at 61–63. At this high level of generality, the dissent insists section 165.540 does not prevent Project Veritas from engaging in investigative journalism of some sort. And it claims that we previously held that restricting unannounced recording does not foreclose the medium of investigative journalism. See

PROJECT VERITAS V. SCHMIDT 39

Dietemann v. Time, Inc., 449 F.2d 245 (9th Cir. 1971). Dissent at 55–56, 62 & n.11, 65. We disagree with this analysis. First, the dissent again fails to recognize the implications of Wasden. Under Wasden, the creation of an unannounced recording of a subject’s unguarded conduct (which would include any statements made in the course of such conduct) is itself a form of protected speech and constitutes “a significant medium” of public expression. 878 F.3d at 1203 (citation and quotation marks omitted). As explained above, section 165.540(1)(c) does not leave ample alternative channels for Project Veritas’s mode of speech. Thus, the dissent’s argument that section 165.540(1)(c) does not foreclose investigative journalism as a journalistic approach misses the mark. At some level of generality, “art” can be made without a paint brush—but neither sculpture nor architecture is a substitute for painting. Moreover, the dissent’s reliance on Dietemann is misplaced. Dissent at 55–56, 62 & n.11, 65. In Dietemann, two journalists used a ruse to gain entry to the plaintiff’s home and then surreptitiously photographed and recorded him without consent. 449 F.2d at 245–46. We held that the plaintiff could state a claim for invasion of privacy under California law because the conduct occurred inside the plaintiff’s home, id. at 248, and because the First Amendment did not “accord newsmen immunity from torts or crimes committed during the course of newsgathering,” id. at 249. But Dietemann has no bearing on the question whether a rule prohibiting unannounced recordings in public places fails to leave open ample alternative channels of communication.

40 PROJECT VERITAS V. SCHMIDT

For this reason, the dissent’s argument that a parade of horribles will result from our analysis—such as the invalidation of “eavesdropping statutes”—is not well-taken. Dissent at 73. As explained, see supra Section III.A., the threshold question is whether the challenged law restricts First Amendment protected speech. Under Wasden, the creation of an unannounced recording is speech protected by the First Amendment. But we are not aware of any cases holding that eavesdropping (without more) is protected speech. Therefore, the First Amendment would not constitute grounds to invalidate a statute prohibiting that conduct. Moreover, we analyzed section 165.540(1)(c) as a prohibition of First Amendment protected speech in public places. See supra Section III.C. Our analysis of the state’s asserted governmental interest and whether its restriction on speech is narrowly tailored would necessarily be different in the context of eavesdropping, where an individual’s heightened privacy interests in his own home are at stake. Nothing we have said today impugns the well-established rule that the First Amendment does not “accord [a speaker] immunity from torts or crimes committed” in service of his speech. Dietemann, 449 F.2d at 249.20 Because we conclude that section 165.540(1)(c) is not a valid time, place, or manner restriction, it cannot be “saved” by striking the two exceptions at issue here. Therefore,

20 The dissent argues that our conclusion that section 165.540(1)(c) is not a valid time, place, or manner restriction, means that the Oregon legislature is “in a catch-22.” Dissent at 69. But a judicial determination that a statute is unconstitutional does not put the legislature in a catch-22 situation; rather, it merely tells the legislature that its enactment has impermissibly infringed on the First Amendment rights of its citizens.

PROJECT VERITAS V. SCHMIDT 41

“severability is not germane.” Bernstein Bros., 661 P.2d at 539. Further, under Outdoor Media Dimensions, we also conclude that the Oregon legislature would not intend the exceptions to be severed, because when Oregon courts analyze severability, they “assum[e] that the legislature prefers to avoid enacting a bill that raises serious questions of constitutionality.” State v. Borowski, 220 P.3d 100, 109 (Or. Ct. App. 2009). If the exceptions were removed, section 165.540(1)(c) would raise serious constitutional issues. This section would prohibit the unannounced recording of police officers performing their official duties or a felony endangering human life. But we have consistently and repeatedly held that “[t]he First Amendment protects the right to photograph and [to] record matters of public interest,” Askins, 899 F.3d at 1044, which includes the right to “observ[e] government[al] operation[s],” Reed v. Lieurance, 863 F.3d 1196, 1211 (9th Cir. 2017), and the commission of a crime, see Obsidian Fin. Grp., LLC v. Cox, 740 F.3d 1284, 1291– 92 (9th Cir. 2014). Requiring a citizen to inform all parties involved to capture governmental officials’ performance of official duties in public places, for example, would substantially impede this speech right by foreclosing a major avenue for citizens to “[g]ather[] information about government officials in a[n unaltered] form that can readily be disseminated to others,” despite the fact that this type of speech “serves a cardinal First Amendment interest in protecting and promoting ‘the free discussion of governmental affairs.’” Glik v. Cunniffe, 655 F.3d 78, 82 (1st Cir. 2011) (citing Mills v. Alabama, 384 U.S. 214, 218 (1966)). Further, an announced recording of a felony in progress would not only tend to reduce the opportunity to capture such evidence, but also tend to imperil the person

42 PROJECT VERITAS V. SCHMIDT

recording. Given the impetus for this exception was to enable police officers to make unannounced recordings of felony drug transactions and felonies endangering human life without first obtaining a court order, see supra pp. 8–9, the legislature would not choose to endanger police by eliminating this exception to the general rule. The dissent suggests that removing the exceptions from the general prohibition in section 165.540(1)(c) would not raise constitutional issues because a court would likely deem section 165.540(1)(c) unconstitutional as applied to an individual who filmed police or other matters of public interest in public places. Dissent at 68–69. But such a conclusion merely acknowledges that the general prohibition itself raises serious constitutional issues. Therefore, severing the exceptions that make the general prohibition content based, and extending the general prohibition to these protected First Amendment activities, would create significant constitutional issues rather than cure them. Under Outdoor Media, we must presume that the Oregon legislature would not retain such a law.21

21 The dissent argues that the legislature would want to retain section 165.540(1)(c) as a stand-alone provision, even if the exception in section 165.540(5)(b) for recording police officers were severed, because the general prohibition in section 165.540(1)(c) “was freestanding for fifty- six years before the legislature adopted the exception that allows the recording of law enforcement officers performing official duties in public.” Dissent at 51; see also Dissent at 67–68. This evinces a misunderstanding of the relevant legislative history. The legislature adopted section 165.540(1)(c) long before Fordyce made clear that such a general prohibition on filming matters of public concern raises serious constitutional questions. See 55 F.3d at 439. Following Fordyce and subsequent opinions reiterating this rule, the legislature added the exception in section 165.540(5)(b)—likely to eliminate this constitutional concern. (Unfortunately, the addition of this exception

PROJECT VERITAS V. SCHMIDT 43

*** Reading section 165.540(1)(c) as a whole, we conclude that it is a content-based speech restriction that cannot survive strict scrutiny because Oregon has not asserted a compelling government interest and because the statute is not narrowly tailored. The statute is also not a valid time, place, or manner restriction because it does not leave open ample alternative channels for communication. Applying Oregon law, we may not sever the exceptions because severing them would not render section 165.540(1)(c) constitutional. Accordingly, we conclude that the statute is facially unconstitutional. REVERSED and REMANDED.

CHRISTEN, Circuit Judge, dissenting: “The right to speak and publish does not carry with it the unrestrained right to gather information.” Zemel v. Rusk, 381 U.S. ^[1], 17 (1965). When it adopted Oregon Revised Statutes section 165.540(1)(c), the Oregon legislature required that notice must be given before in-person oral conversations may be recorded. With this statute, the legislature ensured

rendered section 165.540 a content-based speech restriction, which created a different First Amendment issue.) Given that the Oregon legislature already evinced its intent to avoid the constitutional questions raised when section 165.540(1)(c) was a standalone provision, we must conclude that the legislature would not sever the exception in section 165.540(5)(b), which would merely bring back to life the same constitutional issue that the Oregon legislature faced prior to enacting this exception.

44 PROJECT VERITAS V. SCHMIDT

that Oregonians would be free to engage in the “uninhibited exchange of ideas and information,” 1 without fear that their words could be broadcast beyond their intended audience, appear on the evening news, or worse, be manipulated and shared across the internet devoid of relevant context. Project Veritas engages in undercover investigative journalism, and it finds Oregon’s protection against the secret recording of oral conversations a hindrance to its operations. Project Veritas seeks a ruling declaring section 165.540(1)(c) unconstitutional, arguing there is no distinction between hearing a conversation and secretly recording it. Because the majority does not dispute that the State has a significant interest in protecting the privacy of Oregonians who engage in conversations without notice that their comments are being recorded, our court’s analysis should be straightforward. First, principles of federalism require that we begin from a premise of reluctance to strike down a state statute. Next, following Supreme Court precedent, we should sever the two statutory exceptions that Project Veritas challenges, apply intermediate scrutiny to the content-neutral remainder, recognize that the statute is well- tailored to meet Oregon’s significant interest, and uphold section 165.540(1)(c) as a reasonable time, place, or manner restriction. The majority takes a very different path. It begins by straining to avoid the conclusion that the two exceptions to section 540(1)(c)’s notice requirement that Project Veritas challenges are severable. From there, the majority concludes that severance is inappropriate by implausibly speculating that the Oregon legislature—which the majority faults for

1 Bartnicki v. Vopper, 532 U.S. 514, 532 (2001) (citation omitted).

PROJECT VERITAS V. SCHMIDT 45

overzealously protecting privacy—would have preferred to jettison all of section 540(1)(c) rather than striking the two exceptions. My colleagues do not contest that Oregon has a significant interest in protecting people from unannounced recordings of in-person conversations, but they rewrite the State’s articulated purpose. The purpose Oregon advances is its significant interest in protecting participants from having their oral conversations recorded without their knowledge. The majority recasts the State’s interest as one in “protecting people’s conversational privacy from the speech of other individuals.” Slip Op. at 25. (emphasis added). That reframing of the legislature’s purpose serves as the springboard for the majority’s reliance on an inapplicable line of Supreme Court authority that pertains to state action aimed at protecting people from unwanted commercial or political speech; not protection from speech- gathering activities like Project Veritas’s, which are qualitatively different because they appropriate the speech of others. The majority glosses over this important distinction, and in the end, it declares that all of section 165.540(1)(c) is unconstitutional by concluding that the State’s ban on unannounced recordings leaves no adequate alternative channel of communication. This final rationale is contrary to the reasoning of our own court, which has explained that “hidden mechanical contrivances are [not] ‘indispensable tools’ of newsgathering. Investigative reporting is an ancient art; its successful practice long antecedes the invention of miniature cameras and electronic devices.” Dietemann v. Time, Inc., 449 F.2d 245, 249 (9th Cir. 1971). Because modern technology now allows voice recordings to be manipulated and disseminated worldwide with a few

46 PROJECT VERITAS V. SCHMIDT

keystrokes and clicks, the protection afforded by section 165.540(1)(c) is more important than ever. For all these reasons, I respectfully dissent. I. In 1955, the Oregon legislature enacted what is now section 165.540 of the Oregon Revised Statutes, a wiretapping law that requires the consent of one party before a telecommunication or a radio communication may be recorded in Oregon. See State v. Lissy, 747 P.2d 345, 350 (1987); Or. Rev. Stat. § 165.540(1)(a) (1955). 2 The legislature amended section 165.540 in 1959 to require that anyone wishing to record an in-person conversation must first specifically inform all participants. 3 Lissy, 747 P.2d at 350 & n.4. “[T]he primary concern underlying [§] 165.540(1)(c) was the protection of participants in conversations from being recorded without their knowledge.” State v. Neff, 265 P.3d 62, 66 (Or. Ct. App. 2011). The 1959 amendment was codified as