Amicus Curiae Brief — NetChoice, LLC, dba NetChoice, et al., Petitioners v. Ashley Moody, Attorney General of Florida, et al.
Supreme Court briefNov 22, 2022
Ask Donna
What actually matters in this document.
Text
No. 22-393
In the Supreme Court of the United States
NETCHOICE, LLC, and the COMPUTER &
COMMUNICATIONS INDUSTRY ASSOCIATION,
Cross-Petitioners,
v.
ASHLEY MOODY, ATTORNEY GENERAL
OF FLORIDA, et al.,
Cross-Respondents.
On Cross-Petition for a Writ of Certiorari to the
United States Court of Appeals for the Eleventh Circuit
Brief of Chamber of Progress; Consumer
Technology Association; Engine Advocacy; Global
Project Against Hate and Extremism; HONR
Network; Information Technology & Innovation
Foundation; Interactive Advertising Bureau; IP
Justice; LGBT Tech; Multicultural Media, Telecom
and Internet Council; Safeguarding American
Values for Everyone; The Software & Information
Industry Association; Stop Child Predators;
TechNet; and Washington Center for Technology
Policy Inclusion as Amici Curiae Supporting
Cross-Petitioners
PAUL W. HUGHES
Counsel of Record
SARAH P. HOGARTH
ANDREW A. LYONS-BERG
ALEX C. BOOTA
McDermott Will & Emery LLP
500 North Capitol Street NW
Washington, DC 20001
(202) 756-8000
phughes@mwe.com
Counsel for Amici Curiae
i
TABLE OF CONTENTS
Table of Authorities.................................................... ii
Introduction and Interest of the Amici Curiae .........1
Summary of Argument................................................3
Argument .....................................................................6
I. Platforms rely upon content moderation to
create safe, productive, and enjoyable
environments. ........................................................6
II. S.B. 7072 decimates platforms’ ability to
effectively and usefully curate content. .............. 14
A. The Act forces disgraceful and wasteful
speech onto platform users............................. 15
B. The Act’s onerous reporting requirements
would make content moderation
functionally impossible. .................................. 21
C. The Act compels disclosure of monitoring
methods to malicious actors. .......................... 22
Conclusion ................................................................. 25
Appendix: List of Amici ............................................. 1a
ii
TABLE OF AUTHORITIES
Cases
Ashcroft v. Free Speech Coalition,
535 U.S. 234 (2002) .............................................. 19
New York v. Ferber,
458 U.S. 747 (1982) .............................................. 19
Stratton Oakmont, Inc. v. Prodigy Servs. Co.,
1995 WL 323710 (N.Y. Sup. Ct. 1995) .................. 6
Virginia v. Black,
538 U.S. 344 (2003) .............................................. 19
Statutes
17 U.S.C. § 512 .......................................................... 18
Fla. Stat.
§ 106.011(3)(e) ...................................................... 16
§ 501.2041(1)(b) .................................... 4, 15, 17, 18
§ 501.2041(1)(d) .......................................... 4, 15, 16
§ 501.2041(2)(b) ................................................ 5, 17
§ 501.2041(2)(d) ................................................ 5, 21
§ 501.2041(3)(d) .......................................... 5, 21, 22
§ 501.2041(2)(h)...................................................... 5
§ 501.2041 (2)(j) ................................................ 4, 15
§ 501.2041(e) ........................................................ 16
§ 501.2041(f) ................................................... 16, 17
§ 501.2041(h) ........................................................ 16
§ 501.2041(j) ........................................................ 19
§ 501.2041(6) .................................................... 5, 17
§ 501.2041(7) .................................................... 5, 17
§ 847.001(12) ........................................................ 19
iii
Other Authorities
AARP, Older Americans’ Cybercrime Losses
Soared to $3 Billion in 2021 (Apr. 5, 2022)......... 20
Anti-Defamation League, Online Hate
and Harassment: The American
Experience 2021 § 8 (2021) .................................. 19
Associated Press, Alex Jones’s Audience
and Infowars’ Revenue Grew as
Jones Alleged Sandy Hook School
Massacre Was a Hoax .......................................... 16
Attorney General of Texas, Senior Scams................ 20
Brooke Auxier, About One-in-Five
Americans Who Have Been Harassed
Online Say It Was Because of Their
Religion, Pew Research Center
(Feb. 1, 2021) ........................................................ 11
Shannon Bond, Facebook Takes Down
China-based Network Spreading
False Covid-19 Claims Nat’l Pub.
Radio (Dec. 1, 2021) ............................................. 11
Shannon Bond, Facebook, YouTube and
Twitter Remove Disinformation
Targeting Ukraine, Nat’l Pub. Radio
(Feb. 28, 2022) ...................................................... 10
Sophie Bertazzo, Online Harassment Isn’t
Growing—But It’s Getting More Severe
Trust Magazine (June 28, 2021).......................... 19
Cong. Research Serv., The First Amendment:
Categories of Speech (2019) ................................. 19
iv
Other Authorities—continued
Gabriel J.X. Dance & Michael H. Keller,
Tech Companies Detect a Surge in Online
Videos of Child Sexual Abuse, N.Y. Times
(Feb. 7, 2020) ........................................................ 18
Facebook, Rights Manager ........................................ 18
Facebook, Terms of Use (Nov. 15, 2007) ..................... 8
Google, Transparency Report: YouTube
Community Guidelines Enforcement............. 13, 23
Taylor Hatmaker, Instagram Warns Users
Who Share Russian State Media, Hides
Following Lists in Russia and Ukraine,
TechCrunch (Mar. 8, 2022) .................................. 10
Jean Whitmore, The Arms Race of Models:
Complexify or Die (June 24, 2021)....................... 23
LinkedIn, Professional Community Policies .............. 8
LinkedIn, User Agreement ¶¶ 2.5, 3.4
(Feb. 1, 2022) .......................................................... 9
Meta, Facebook Community Standards:
Account Integrity and Authentic Identity ............ 12
Meta, Facebook Community Standards:
Report: Adult Nudity and Sexual Activity ......... 23
Meta, Facebook Community Standards: Child
Sexual Exploitation, Abuse and Nudity .............. 12
Meta, Facebook Community Standards
Enforcement Report: Fake Accounts
(Feb. 2022) ............................................................ 12
Meta, Facebook Community Standards:
Hate Speech ......................................................... 19
v
Other Authorities—continued
Meta, Facebook Community Standards:
Spam............................................................... 10, 22
Meta, Facebook Community Standards:
Violent and Graphic Content ............................... 11
Meta, Meta Reports First Quarter 2022
Results (Apr. 27, 2022) ......................................... 12
Meta, Removing Holocaust Denial
Content (Oct. 12, 2020) ........................................ 11
Nat’l Center for Missing & Exploited
Children, CyberTipline 2021 Report ................... 18
Jack Nassetta & Kimberly Gross, State Media
Warning Labels Can Counteract the Effects of
Foreign Misinformation, Harvard Kennedy
School Misinformation Review (Oct. 30, 2020) ... 16
Reddit, Is Posting Someone’s Private or Personal
Information Okay? (July 13, 2020) ...................... 12
Reddit, Never Post Intimate or Sexually
Explicit Media of Someone Without
Their Consent (Mar. 7, 2022) ............................... 12
Reddit, Reddit Content Policy ................................... 13
Amanda Silberling, Facebook Grilled in
Senate Hearing over Teen Mental
Health, TechCrunch (Sept. 30, 2021) .................. 11
Michelle Shen, Twitter Accounts Tied to China
Lied that Covid Came from Maine Lobsters,
USA Today (Oct. 22, 2021) .................................. 10
vi
Other Authorities—continued
Steven Sumner et al., Temporal and
Geographic Patterns of Social Media
Posts About an Emerging Suicide
Game, J. Adolescent Health (2019) ..................... 20
Twitter, About Government and State-Affilliated
Media Account Labels on Twitter ........................ 16
Twitter, Abusive Behavior ........................................ 19
Twitter, Hateful Conduct Policy ............................... 11
Twitter, Non-Consensual Nudity Policy
(Nov. 2019) ........................................................... 12
Twitter, Our Range of Enforcement Options ............ 13
Twitter, Platform Manipulation (Jan. 25, 2022) ..... 13
Twitter, Platform Manipulation (July 28, 2022) ..... 23
Twitter, Rules Enforcement (July 28, 2022) ............. 23
Emily A. Vogels, The State of Online Harassment,
Pew Research Center (Jan. 13, 2021) ................. 21
YouTube, Child Safety Policy ............................. 12, 19
YouTube, Spam, Deceptive Practices, &
Scams Policies ...................................................... 10
1
INTRODUCTION AND INTEREST OF THE
AMICI CURIAE 1
The advent of the Internet created powerful new
means for people to gather socially, providing us novel
ways to share our stories, to mourn and celebrate together, and to discuss—indeed, often debate—topics
ranging the entire breadth of human experience.
Online platforms allow friends and family not just to
keep in touch, but to maintain lively conversations
across time and distance. These communities have
created millions of new relationships—from the professional to the romantic. Countless businesses have
been built with these platforms. Many people have
found celebrity online, be it for a popular video series,
a newly discovered musical talent, or a simple tweet
gone viral. And far more of us consume this content,
enjoying the creative talents of others. We turn to social media for entertainment, for news, for education,
and for self-expression.
The power of these platforms derives from their
immediacy, interactivity, and diversity of content,
driven by users who not just create content, but also
react, share, and comment. Yet the very features that
make platforms compelling also introduce risks of
misuse that, if unchecked, can overrun services. Absent basic content moderation, unsolicited commercial
advertisements—often referred to as “spam” in Internet vernacular—will quickly overtake a platform,
drowning out the valuable content. And that is just
the start. Vile, graphic videos and images can
All parties have consented to the filing of this brief via blanket
letters of consent. No counsel for a party authored this brief in
whole or in part, and no person other than the amici, their members, or their counsel has made a monetary contribution intended
to fund the preparation or submission of the brief.
1
2
proliferate. Reprehensible speech—such as the glorification of terrorist attacks and the dissemination of
pro-Nazi views—will target communities struggling
to recover from violence and loss. Hostile foreign governments may hijack these platforms to engage in disinformation campaigns on American soil.
Recognizing these realities, online platforms exercise editorial controls to select, arrange, and curate
content consistent with the messages and experiences
they wish to promote. While the precise rules differ
across sites according to their purposes and values,
these platforms broadly seek to create experiences for
their users that are safe, productive, and enjoyable.
Florida Senate Bill 7072 (S.B. 7072) threatens to
eviscerate this content moderation. For only those
online platforms that are the most successful—those
with 100 million or more regular users, or gross revenues of over $100 million—Florida prohibits content
moderation outright in many cases, and attaches the
potential for $100,000 in statutory damages per occurrence in any other case, if a judge later decides that
the action was “[in]consistent” with other contentmoderation actions taken by the platform. For platforms that take millions, if not billions, of contentmoderation actions per year, this liability is guaranteed to chill platforms’ essential expressive activity.
As cross-petitioners have demonstrated, and as
the Eleventh Circuit largely held, the First Amendment does not tolerate such enormous infringement of
the rights of the owners and operators of these platforms. These are privately owned spaces, and the platform operators may employ content moderation rules
to curate the speech and expressive communities they
wish to promote.
3
Amici focus on the enormous practical consequences that would result if S.B. 7072—and other,
similar laws enacted or under consideration elsewhere—is allowed to take effect. At present, platforms
rely on content moderation to help millions of Americans work, play, learn, shop, connect, and express
themselves through online platforms free from harassment, disinformation, and incendiary content. But,
by foreclosing any meaningful ability of platforms to
engage in content curation, S.B. 7072 erases their
very utility, denies them editorial control over the
speech and ideas they host, and threatens severe
harm to platforms and to their users. It also requires
platforms to redesign their essential operations and to
implement onerous notice processes for each of the billions of pieces of content moderated.
Amici are organizations that are all deeply interested in ensuring that Americans may participate in
healthy online environments. Amici and their members thus have a strong interest in ensuring that Florida’s S.B. 7072, and other similar laws around the
country, are not permitted to threaten, disrupt, or destroy vibrant and diverse online communities.
SUMMARY OF ARGUMENT
The Court should grant the conditional cross-petition, along with Florida’s petition in No. 22-277, to review the entirety of S.B. 7072 and confirm that the
First Amendment does not tolerate such restrictions
and burdens on the speech of private actors.
I. Any entity that hosts content of another—be it
a bookstore, a magazine, or an online platform—engages in core speech activities when it chooses what
content to present and how to display it. In establishing rules and procedures for content selection and
moderation, these entities embed the environments
4
they create with their unique values and perspectives,
giving voice to the platform’s own decisions as to what
content warrants promotion.
Content curation is nothing new—in fact, quite
the opposite. Platforms that distribute third parties’
messages have always done so. Bookstores choose
which books to stock and to promote; newspapers
choose which letters to the editor to publish. Online
platforms are no different: They choose what types of
user-generated content fit the platform—whether targeted to social interactions (like Facebook), video
sharing (like YouTube), character-limited conversation (like Twitter), professional networking (like
LinkedIn), travel reviews (like Tripadvisor), or product reviews (like Amazon). As part of their contentcuration choices, platforms must also choose what
content not to distribute. Platforms have long used
terms and conditions and policy statements to advise
users of what content is appropriate for the platform
and what content is not. Through these processes,
platforms engage in their own expression and protect
their communities—by removing or limiting spam,
scams, fraud, phishing, fake accounts, state-sponsored misinformation, hate speech, violence, threats,
and abusive and offensive content.
II. S.B. 7072 would effectively gut platforms’ efforts at content moderation, to the ultimate detriment
of their users. With a narrow exception for technical
obscenity, the law flatly prohibits platforms from
“censor[ing]”—to include removing, age-gating, or
even posting a counter-speech disclaimer—the content of any organization with a sufficiently large audience, thus enlisting platforms to serve up to their
users the messages of hostile foreign governments and
conspiracy theorists, among others. Fla. Stat.
§ 501.2041(1)(b), (1)(d), (2)(j). It also contains a similar
5
prohibition on moderating content by, or about, any
political candidate—and it does not take much imagination to conceive of content “about” a political candidate that is vile, hateful, threatening, and unwanted
by a platform’s users. Id. § 501.2041(2)(h). And the
law creates a private right of action, with $100,000 in
statutory damages per instance, for any user whose
content has been moderated in an “[in]consistent”
manner compared to the platform’s past moderation
actions, creating enormous, chilling litigation exposure for platforms that take many millions, if not billions, of content-moderation actions each year. Id.
§ 501.2041(2)(b), (6), (7).
Apart from its substantive prohibitions, S.B. 7072
also burdens the expressive activities of platforms
through onerous disclosure requirements. The law
mandates that platforms provide a “thorough rationale” to the user, within seven days, for each and
every one of its millions or billions of content-moderation actions each year, imposing a compliance obligation that will be practically impossible to meet. Fla.
Stat. § 501.2041(2)(d), (3). And it goes even further,
requiring platforms to disclose “a precise and thorough explanation” of its content-detection “algorithms” to the poster of deleted content. Ibid. That is,
platforms must hand detailed information about their
content moderation tools to the very actors attempting
to evade them, crippling platforms in their constantly
evolving fight against these malicious actors, to the
ultimate detriment of the users—everyday Americans—that the platforms are attempting to serve.
6
ARGUMENT
I.
Platforms rely upon content moderation to
create safe, productive, and enjoyable
environments.
Successful online platforms create digital spaces
that their users wish to inhabit. Some are professional
networks that help create new business opportunities.
Other platforms allow us to share our daily thoughts
in short bursts, from the profound to the silly. Yet
other platforms allow us to upload and share videos or
images that we create. These platforms have given life
to an array of online communities, allowing us to interact in a variety of meaningful ways.
From the beginning, while allowing their users to
contribute their own, individual content, platforms
have formed rules governing what content they will
allow. See Stratton Oakmont, Inc. v. Prodigy Servs.
Co., 1995 WL 323710, at *2 (N.Y. Sup. Ct. 1995) (describing “content guidelines” of an early online finance
message board). And, for the content they permit,
platforms choose how that material is displayed to users.
These content curation policies are hardly surprising, as platform owners have always chosen what content they will disseminate and the manner in which
they will organize their content for users. Magazines
and newspapers choose what stories they will publish,
and where in the publication they will appear. They
likewise choose which reader-generated letters they
will print, and how and when they will print them.
Bookstores and newsstands choose the material they
will offer; some of their wares will be prominently displayed in the front, while other material will be kept
in the rear, tucked away on a bottom shelf.
7
Platforms of all sorts curate the content they offer
to create the specific environment that they intend to
offer, consistent with the platform’s key values. Just
as a bookstore will organize its contents to be readily
accessible to its customers, even placing some books
in the front window to draw in passers-by, online platforms arrange content to entice, inform, help, and
even delight their users. Some platforms organize content topically, like Reddit’s “subreddits” devoted to
particular subjects. Others make recommendations,
like Facebook’s “suggested for you” posts. Yet others
highlight popular content, like Twitter’s “trending”
feature. These features create the very essence of the
online communities.
In fact, the enormous volume of material posted
online enhances the need for platforms to curate content. Users generally engage with platforms because
there is content that they wish to see—be it breaking
news or posts from their close friends or discussions
on a topic of particular interest. Many platforms have
mechanisms to individually prioritize content for each
of their millions of users, finding individualized harmonies among the din of otherwise irrelevant posts.
These tools are one essential form of content moderation.
Choices about what content to distribute necessarily require choices about what content not to distribute. It is unremarkable that a children’s bookstore
would not stock romance novels or that a home-andgarden television network would not air a baseball
game. No shopper or viewer would be surprised. The
same goes for online platforms. Choices about the content that a platform will not distribute are just as crucial in creating and preserving the platform’s desired
environment. Take, for example, YouTube. Along with
its flagship online video-sharing site, the company
8
offers a separate application and website called
YouTube Kids with content specifically curated for
children. The content available on each is, of course,
markedly different—and expectedly so. Those differences are the direct result of YouTube engaging in
careful content moderation tailored for different audiences.
Content moderation thus empowers platforms to
foster online communities that fit the specific needs
and values of that particular platform and its users
and advertisers. To do so, platforms have long included standard terms and conditions limiting the
content that may be published on their platforms. And
users have long agreed to and accepted these terms.
Consider Facebook. It began as a social community for college students and, by September 2006, for
anyone over 13 years of age. See Meta, Our History,
about.facebook.com/company-info. Facebook’s November 2007 terms of service barred users from posting
“any content that we deem to be harmful, threatening,
unlawful, defamatory, infringing, abusive, inflammatory, harassing, vulgar, obscene, fraudulent, invasive
of privacy or publicity rights, hateful, or racially, ethnically or otherwise objectionable” and warned that it
might remove violative content and terminate or bar
membership. Facebook, Terms of Use 2-3, 8 (Nov. 15,
2007), perma.cc/T65S-63LW. Through these terms,
Facebook made clear the type of community it hoped
to foster. As its popularity confirms, users and advertisers liked the community that Facebook created.
LinkedIn has a purposefully different flavor. It is
intended for professional connections. LinkedIn policy
requires that content be kept “professionally relevant”
as well as safe and trustworthy. LinkedIn, Professional Community Policies (visited November 4,
2022), perma.cc/V4NX-TUDZ. Violating these policies
9
can result in a user’s account being restricted, suspended, or terminated, and LinkedIn warns that it
need not publish any particular content and can remove it with or without notice. LinkedIn, User Agreement ¶¶ 2.5, 3.4 (Feb. 1, 2022), perma.cc/B8EK-5LCX.
Tripadvisor has different terms tailored to its
community’s needs. Tripadvisor specifically prohibits
posting comments “that are not relevant to travel or
incite non-travel related discussions” and warns that
such content will be removed. Tripadvisor, Content &
Community Guidelines (May 4, 2022), perma.cc/Y529LEPL.
Amazon similarly has policies keeping its product
reviews “helpful, relevant, meaningful, and appropriate.” Amazon, Community Guidelines (visited November 4, 2022), perma.cc/SLZ7-LCT8. Amazon makes
clear that violating the guidelines threatens the trustworthiness, safety, and utility of its community, and
that content may be removed or accounts suspended
or terminated for violating them. Id.
Through enforcing these policies, platforms foster
and protect the purpose of their respective communities. Particularly in the online context, the ease of
posting and reproducing content means that, absent
content moderation policies, platforms would be overrun with lawful but nonetheless awful, irrelevant,
low-quality, and nuisance content. Platforms thus use
their policies to preclude activity that, while not in violation of the law, is destructive of the platform itself.
These harmful activities include content like
spam, phishing, scams, and deceptive practices. For
example, YouTube bars video spam—including videos
that are excessively posted, that promote get-richquick schemes, or that direct viewers to sites spreading harmful software, among many other things. See
10
YouTube, Spam, Deceptive Practices, & Scams Policies (visited November 4, 2022), perma.cc/NR5HZMPJ. YouTube also bars spam in its comments section, precluding users from leaving large amounts of
identical or repetitive comments, comments intended
to gather personal information from viewers, or comments that misleadingly direct viewers to other sites.
Ibid. Facebook similarly has policies designed to protect against spam; for example, it bars posting content
at very high frequencies, posts that direct users off the
site through misleading links, or messages that offer
false or non-existent services or functionality. Meta,
Facebook Community Standards: Spam (visited November 4, 2022), perma.cc/WD7C-ZTVF.
Platforms also moderate extreme and misleading
messaging—including pro-Nazi speech, support of terrorism, and misinformation campaigns. Recently,
platforms have aggressively removed disinformation
intended to skew public perceptions about the Russian invasion of Ukraine. See, e.g., Shannon Bond, Facebook, YouTube and Twitter Remove Disinformation
Targeting Ukraine, Nat’l Pub. Radio (Feb. 28, 2022),
perma.cc/CM5Q-P9KT. These efforts have included
stemming the flow of state-backed campaigns by using warning labels and downranking Russian statemedia propaganda pertaining to the Ukraine invasion. See, e.g., Taylor Hatmaker, Instagram Warns
Users Who Share Russian State Media, Hides Following Lists in Russia and Ukraine, TechCrunch (Mar. 8,
2022), perma.cc/4QFU-EVX5.
Similar actions have been taken against proChina organizations spreading disinformation about
the origins of the Covid-19 pandemic. For example,
Twitter removed accounts contending that the Covid19 virus was spread to China by the U.S. military or
by a shipment of Maine lobsters. See Michelle Shen,
11
Twitter Accounts Tied to China Lied that Covid Came
from Maine Lobsters, USA Today (Oct. 22, 2021),
perma.cc/9BCF-B2YN. Facebook removed more than
600 accounts, pages, and groups from a Chinese influence organization spreading posts alleging that the
United States was pressuring the World Health Organization to blame China for the Covid-19 virus.
Shannon Bond, Facebook Takes Down China-based
Network Spreading False Covid-19 Claims, Nat’l Pub.
Radio (Dec. 1, 2021), perma.cc/SJ5A-NM48.
Platforms also moderate hate speech, like content
that denies or distorts the Holocaust, espouses a hope
that members of a certain nationality would all die, or
that praises harassment of members of a particular
religion. See Meta, Removing Holocaust Denial Content (Oct. 12, 2020), perma.cc/WWW4-BCLZ; Twitter,
Hateful Conduct Policy (visited November 4, 2022),
perma.cc/ZBT4-PJMT; Brooke Auxier, About One-inFive Americans Who Have Been Harassed Online Say
It Was Because of Their Religion, Pew Research Center (Feb. 1, 2021), perma.cc/T3UT-AQHV.
Platforms also may restrict content promoting
self-harm, especially that directed to minors. Indeed,
federal lawmakers on both sides of the aisle have
urged platforms to take steps to help support teen users’ mental health. Amanda Silberling, Facebook
Grilled in Senate Hearing over Teen Mental Health,
TechCrunch (Sept. 30, 2021), perma.cc/PN23-M5E6.
Platforms accordingly moderate an array of content
that, although not unlawful, could nonetheless be extremely dangerous for impressionable users, including content promoting eating disorders, practices of
self-mutilation, and even suicide.
Similarly, content glorifying violence or the suffering of others may be removed or subject to a content
warning. See Meta, Facebook Community Standards:
12
Violent and Graphic Content (visited November 4,
2022), perma.cc/T8ZU-RDZD. Facebook bars nonmedical videos of dismemberment, visible internal organs, human burning, and throat-slitting, but applies
content warnings and age-gates to videos in the medical setting or to photos depicting these same topics.
Facebook likewise applies content warnings and agegates to content involving animal cruelty or abuse.
Platforms moderate content that attempts to
groom minors for sexual encounters, celebrates child
sexual abuse, and involves the non-consensual posting of another’s intimate images. 2 Platforms may limit
publication of private information of users, or content
that impersonates others. 3 This is all just a small sampling of the range of content subject to moderation.
Achieving content moderation at the scale of these
online platforms is a herculean task. Platforms remove millions or even billions of pieces of content for
violating policies every year. To put it in context, in
the fourth quarter of 2021, Facebook removed 1.7 billion fake accounts—nearly the same number as it has
active daily users. 4 Over the same period, YouTube
See Meta, Facebook Community Standards: Child Sexual Exploitation, Abuse and Nudity (visited November 4, 2022),
perma.cc/F5LM-MXDV; YouTube, Child Safety Policy (visited
November 4, 2022), perma.cc/5SHG-5DNM; Twitter, Non-Consensual Nudity Policy (Nov. 2019), perma.cc/52DA-YR7R; Reddit, Never Post Intimate or Sexually Explicit Media of Someone
Without Their Consent (Mar. 7, 2022), perma.cc/Y94G-CZFC.
2
See Reddit, Is Posting Someone’s Private or Personal Information Okay? (July 13, 2020), perma.cc/F5FP-CCF7; Meta, Facebook Community Standards: Account Integrity and Authentic
Identity (visited November 4, 2022), perma.cc/HF9T-V5QH.
3
See Meta, Facebook Community Standards Enforcement Report: Fake Accounts (Feb. 2022), tinyurl.com/3xxe8vue; Meta,
4
13
removed more than 1.26 billion comments, 99.5% of
which were spam, and more than 3.8 million channels,
89% of which were spam. 5 Twitter suspends over half
a million spam accounts per day 6—more than 130 million in the first half of 2021. 7 As these numbers reflect,
content moderation is essential to a platform’s ability
to provide the community that it desires.
Platforms thus use carefully calibrated moderation policies to build the environments they wish to
make available to users. These moderation decisions
cut across an enormously broad range of topics. And
there are a wide variety of moderation tools available,
including content prioritization or demotion, suspension or removal of entire accounts, removal of certain
content, age-gating, and affixing labels or warnings,
among a host of other techniques. 8 There is simply no
one-size-fits-all solution for content moderation. The
variety of approaches taken by online platforms only
serves to confirm that decisions about what content to
distribute—and how to display it—constitute core expressive activity.
Meta Reports First Quarter 2022 Results (Apr. 27, 2022),
perma.cc/3V4M-FS6H.
Google, Transparency Report: YouTube Community Guidelines
Enforcement (visited November 4, 2022), tinyurl.com/5cp5npev.
5
Parag Agrawal (@paraga), Twitter (May 16, 2022 12:26 PM),
perma.cc/33VK-46B3.
6
Twitter,
Platform
perma.cc/2ZQL-LMPR.
7
Manipulation
(Jan.
25,
2022),
Reddit, Reddit Content Policy (visited November 4, 2022),
perma.cc/XMD7-92SF (describing “variety of ways of enforcing
our rules”); Twitter, Our Range of Enforcement Options (visited
November 4, 2022), https://perma.cc/R48Y-HG5Y (describing the
different levels of enforcement actions, ranging from “Tweetlevel” to “Account-level” enforcement).
8
14
In all, platforms craft their services to provide environments and experiences that reflect their values—including by creating safe online spaces for their
audiences. Platforms remove content glorifying the
most horrific violence, from school shootings to racially motivated murders. They work to eliminate disinformation campaigns, including those sponsored by
foreign powers. And they seek to foster environments
safe for children, working to prevent content that
could cause grave injury via sexual exploitation or
self-harm. In making these choices, platforms necessarily speak a message to all who choose to listen.
II. S.B. 7072 decimates platforms’ ability to
effectively and usefully curate content.
For the select few platforms to which it applies,
S.B. 7072 upends most all of the content-moderation
policies that providers and users rely upon to structure their online communities. It does so through
three principal provisions: It prohibits so-called “censor[ship]” altogether in many cases and attaches the
threat of crippling liability in all others; it requires individualized notice-and-appeal procedures to accompany the millions—if not billions—of moderation actions taken by platforms each year; and it orders disclosure of the tools that platforms use to restrict the
proliferation of malicious content.
Collectively, these requirements essentially eliminate platforms’ ability to meaningfully moderate the
content they display to their users. The consequences
of such a policy would be dire: Hate speech would prevail online unchecked, disinformation would bombard
users, and children would be exposed to horrifically
dangerous content. The law must be entirely repudiated.
15
A. The Act forces disgraceful and wasteful
speech onto platform users.
S.B. 7072 contains several provisions that either
directly prohibit or unduly burden platforms’ contentmoderation activities.
1. Most obvious is the law’s flat prohibition on
“censor[ing]”—to include “delet[ing], regulat[ing], restrict[ing], edit[ing], alter[ing], * * * or post[ing] an
addendum to—the content of a “journalistic enterprise based on the content of its publication or broadcast,” with a narrow exception for content meeting the
technical definition of obscenity. Fla. Stat.
§ 501.2041(1)(b), (2)(j). “[J]ournalistic enterprise” is
defined solely by an entity’s volume of content and
number of users (see id. § 501.2041(1)(d)), making this
prohibition quite sweeping. As the Eleventh Circuit
observed:
[T]he provision is so broad that it would prohibit a child friendly platform like YouTube
Kids from removing—or even adding an age
gate to—soft-core pornography posted by
PornHub, which qualifies as a “journalistic
enterprise” because it posts more than 100
hours of video and has more than 100 million
viewers per year.
Pet. App. 62a. Similarly, “S.B. 7072 would seemingly
prohibit Facebook or Twitter from removing a video of
a mass shooter’s killing spree if it happened to be reposted by an entity that qualifies for ‘journalistic enterprise’ status.” Id. at 62a n.23. And it would force
platforms to host and disseminate without disclaimers, for example, the lies of conspiracy theorist Alex
16
Jones that the Sandy Hook school shooting was a
hoax. 9
Less sensationally, but perhaps even more ominously, the prohibition on moderating content posted
by “journalistic enterprise[s]” would surely be seized
upon by foreign government-controlled media outlets
to spread disinformation and intentionally sow discord in the United States, and platforms would be
forced to cease their current efforts to prevent these
harms. 10
S.B. 7072 also flatly prohibits content-moderation
activities—specifically, prioritizing or deprioritizing
content, and “limit[ing] or eliminat[ing] the exposure
of a user or content”—with respect to “content and material posted by or about” candidates for political office. Fla. Stat. § 501.2041(e), (f), (h) (emphasis added).
Not only does that provision give a free pass to anyone
wishing to enlist platforms in disseminating objectionable material, simply by registering as a political
candidate (id. §§ 501.2041(h), 106.011(3)(e)), but its
application to any content “about” a political candidate would seem to mean that platforms would be
Compare Associated Press, Alex Jones’s Audience and Infowars’ Revenue Grew as Jones Alleged Sandy Hook School Massacre Was a Hoax, Marketwatch (Sept. 19, 2022) (reporting audience figures for Jones’s site Infowars), perma.cc/5VBX-NALF,
with Fla. Stat. § 501.2041(1)(d) (qualifications for “journalistic
enterprise” status).
9
See, e.g., Twitter, About Government and State-Affilliated Media Account Labels on Twitter (visited Nov. 3, 2022),
perma.cc/K7UD-QSXR; Jack Nassetta & Kimberly Gross, State
Media Warning Labels Can Counteract the Effects of Foreign
Misinformation, Harvard Kennedy School Misinformation Review (Oct. 30, 2020) (reporting results of peer-reviewed study
showing that labeling of state-controlled media can mitigate the
effects of misinformation).
10
17
powerless to “limit” its users’ “exposure” to posts, for
example, advocating violence against a politician (id.
§ 501.2041(f)).
2. Apart from its flat bans on moderating these
broad categories of content, S.B. 7072 also prohibits
platforms from “censor[ing]” content—again, a term
that encompasses essentially any act of content moderation (Fla. Stat. § 501.2041(1)(b)—in a way that a
court later deems was not “consistent * * * among its
users on the platform” (id. § 501.2041(2)(b)). The law
provides no guidance on the meaning of “consistent.”
And it combines this vague prohibition with a private
right of action for enforcement and draconian penalties including statutory damages of up to $100,000 for
each individual act of content moderation, plus actual
damages, punitive damages, and attorney fees in certain cases. Id. § 501.2041(6), (7).
For platforms that moderate millions, if not billions, of pieces of content every year (see pages 12-13,
supra), it is not difficult to see the chilling effect of
$100,000 statutory damages exposure for each and
every one of those decisions, particularly when paired
with a vague and undefined substantive prohibition
on “[in]consistent” content moderation. This crippling
exposure would force platforms to decline to enforce
their policies and standards against all but the most
obvious violations, with the result that users would be
inundated with harmful, vile, and wasteful content,
against both their own wishes and those of the platforms. Or, platforms could take the opposite tack,
finding it too risky to host any user-generated content
that has not been pre-vetted—with the similarly
harmful result that their online worlds will lack
much, if any, content at all.
Indeed, even the removal of obviously illegal content could be chilled by this litigation risk. Platforms
18
may employ automated detection technologies to identify illegal content, intellectual property infringement, or child sexual abuse. Platforms have invested
substantial resources in developing sophisticated automated tools to fight child sexual abuse by quickly
identifying and removing content depicting such abhorrent acts, contributing to the 29.1 million online
provider reports to the National Center for Missing
and Exploited Children of child sexual abuse imagery
and other abuse in 2021. 11 They have also developed
similar tools to fight piracy on a broad scale—tools
like Meta’s Rights Manager and YouTube’s Content
ID. 12 See 17 U.S.C. § 512. These measures protect intellectual property rights millions of times every day.
That said, as sophisticated as these tools are, they
come with small error rates, with each error potentially resulting in litigable claims of “censor[ship]” under this Florida law. S.B. 7072 is thus a strong disincentive for platforms to innovate and collaborate on
improving automatic detection measures for illegal
material, which will harm child victims of abuse and
stymy content owners trying to police their rights in
the Internet era.
3. The reality is that there is a mismatch between
the speech the First Amendment protects from government regulation and the speech that users and
platforms want as part of their online communities.
Gabriel J.X. Dance & Michael H. Keller, Tech Companies Detect a Surge in Online Videos of Child Sexual Abuse, N.Y. Times
(Feb. 7, 2020), perma.cc/CAP3-ARR3; Nat’l Center for Missing &
Exploited Children, CyberTipline 2021 Report (visited November
4, 2022), perma.cc/H3S5-5HND.
11
YouTube, How Content ID Works (visited November 4, 2022),
perma.cc/J22X-DFLT; Facebook, Rights Manager (visited November 4, 2022), rightsmanager.fb.com/.
12
19
Compare Cong. Research Serv., The First Amendment: Categories of Speech (2019), perma.cc/V6KB5UEV, with Anti-Defamation League, Online Hate
and Harassment: The American Experience 2021 § 8
(2021), perma.cc/CBV2-SQ23; Sophie Bertazzo,
Online Harassment Isn’t Growing—But It’s Getting
More Severe, Trust Magazine (June 28, 2021),
perma.cc/2259-2GX7. Platforms moderate—and users
want them to moderate—more broadly than the First
Amendment’s limits on the government’s regulation of
speech, including:
• all manner of threats, 13 not merely “true
threats” unprotected by the First Amendment
(see Virginia v. Black, 538 U.S. 344 (2003));
•
content that dehumanizes or discriminates
against individuals based on perceived membership in a protected class 14;
•
content that sexualizes children, 15 even if it
does not involve actual children (see, e.g., Ashcroft v. Free Speech Coalition, 535 U.S. 234
(2002) (holding ban on virtual child pornography unconstitutional)), does not contain “visual” depictions of children (New York v. Ferber, 458 U.S. 747, 764 (1982)), or does not
meet the technical definition of obscenity (Fla.
Stat. §§ 501.2041(j), 847.001(12)).
The sad truth is that misleading, vile, hateful,
graphic, or offensive content has an outsized ability to
See, e.g., Twitter, Abusive Behavior (visited November 4,
2022), perma.cc/N8Z8-YT3Z.
13
See, e.g., Meta, Facebook Community Standards: Hate Speech
(visited November 4, 2022), perma.cc/4G4V-HXG8.
14
See, e.g., YouTube, Child Safety Policy (visited November 4,
2022), perma.cc/5SHG-5DNM.
15
20
drown out other material online. Under S.B. 7072, the
inevitable result is that platforms will have to choose
between silencing too much content through overbroad but technically “consistent” content moderation
(or, indeed, deciding that it is too risky to host any
content that is not pre-vetted, leading to the same result), or serving up horrific messages and images to
their users. And they will be powerless to stop entities
that meet the lax definition of “journalistic enterprise”—which, again, includes entities like conspiracy
theorists and foreign state-controlled media—from
enlisting the platforms in disseminating whatever
content they wish, no matter how hateful or vile, so
long as it is not technically obscene.
The result is that platforms without any content
(because of overbroad application of content moderation to ensure “consisten[cy]”) or inundated with distasteful content will be effectively unusable for most,
if not all, legitimate users. The latter situation presents particular risk for vulnerable communities like
children who may be exposed to age-inappropriate
content, teens who may be exposed to content encouraging self-harm, 16 the elderly who are targets for
scams, 17 or for groups that are repeatedly and specifically targeted for hate and harassment on account of
Steven Sumner et al., Temporal and Geographic Patterns of
Social Media Posts About an Emerging Suicide Game, 65 J. Adolescent Health 94-100 (2019).
16
Attorney General of Texas, Senior Scams (visited November 4,
2022), perma.cc/XY4S-PAXN (noting scams targeting seniors
“are especially common online”); AARP, Older Americans’ Cybercrime Losses Soared to $3 Billion in 2021 (Apr. 5, 2022),
perma.cc/M5FB-65JM.
17
21
their gender, sex, race, nationality, religious affiliation, or sexual orientation. 18
B. The Act’s onerous reporting requirements
would make content moderation
functionally impossible.
S.B. 7072 also imposes burdensome procedural
obligations. For any broadly defined act of “censor[ship],” “shadow ban[ning],” or “deplatform[ing],”
the platform must “notify[] the user” within seven
days and “[i]nclude a thorough rationale explaining
the reason that the social media platform censored the
user.” Fla. Stat. § 501.2041(2)(d)(1), (3).
Given that some platforms—particularly those
targeted by S.B. 7072—moderate hundreds of millions
if not billions of pieces of content each year, there is
no practical way to create the procedures required by
S.B. 7072 at the necessary scale. Take, for example,
the 1.7 billion fake accounts that Facebook removed
in the fourth quarter of 2021. S.B. 7072 would apparently obligate Facebook to send presumably billions of
notices including “a thorough rationale explaining”
the action, each on a seven-day clock. The same procedures would attach to YouTube, with the 1.26 billion
comments it removed (99.5% of which were spam) in
the fourth quarter of 2021.
These demands exact an enormous toll on platforms, particularly when coupled with S.B. 7072’s timing requirements and multiplied across the many
types of content they remove. This undue burden will
require platforms to divert enormous resources away
from building features and maintaining their
Emily A. Vogels, The State of Online Harassment, Pew Research Center (Jan. 13, 2021), perma.cc/4CZN-5TGU (finding
that 40% of Americans have experienced online harassment).
18
22
communities in ways that users actually want—if the
law’s mandates are even achievable at all. Platforms
will thus again be forced to either forgo content moderation in its entirety or reduce moderation to only a
shadow of the role it currently plays in curating user
experiences. Either way, it is the users who will ultimately lose the services that they currently enjoy.
C. The Act compels disclosure of monitoring
methods to malicious actors.
Apart from the “thorough rationale” requirement,
S.B. 7072’s disclosure provision also mandates that a
platform disclose to the poster of moderated or removed content “a precise and thorough explanation of
how the social media platform became aware of the
censored content or material, including a thorough explanation of the algorithms used, if any, to identify or
flag the user’s content or material as objectionable.”
Fla. Stat. § 501.2041(3)(d) (emphasis added). While
many of the covered platforms and amici organizations support enhanced transparency, this requirement goes well beyond what is necessary for adequate
content moderation.
This disclosure obligation—which essentially requires platforms to self-report the specifics of their
content-filtering methods to the very users posting objectionable content—will have disastrous consequences for keeping spam, terrorist promotion, and
other distasteful content off platforms. This is a problem of monumental import for platforms. Facebook
took action against 1.8 billion spam postings 19 and
16.1 million posts supporting terrorism in Q1 of 2022
Meta, Facebook Community Standards Enforcement Report:
Spam (visited Nov. 3. 2022), tinyurl.com/4trwdkcw.
19
23
alone. 20 Over the same period, YouTube removed more
than 3.5 million spam channels and over 100,000
channels for child safety. 21 In the second half of 2021,
Twitter challenged more than 133 million suspected
spam accounts and took action on more than 940,000
accounts involving abuse or harassment. 22
As it stands now, platform operators must constantly update and refine their processes to block such
content because those who seek to hijack online platforms for their own profit continually evolve their
strategies to evade moderation. As one researcher put
it, “[b]oth the spammers and spam filter-builders are
under pressure to evolve or die.” Jean Whitmore, The
Arms Race of Models: Complexify or Die 9 (June 24,
2021), ssrn.com/abstract=3867464; see also Kevin
Gosschalk, AI vs. AI: The Digital Arms Race to Fight
Fraud (Oct. 24, 2019), tinyurl.com/9f7jb9w6. Twitter’s
former CEO recently explained these concepts, describing how “fighting spam is incredibly dynamic” because “[t]he adversaries, their goals, and tactics
evolve constantly” in response to platforms’ efforts.
Parag Agrawal (@paraga), Twitter.com (May 16,
2021), perma.cc/33VK-46B3.
The algorithm disclosure required by S.B. 7072
would inevitably enable bad actors, spammers, and
scammers to better circumvent content moderation
techniques. These disclosure obligations will cause
Meta, Facebook Community Standards Enforcement Report:
Adult Nudity and Sexual Activity (Visited Nov. 3, 2022), tinyurl.com/3kye4tux.
20
Google, Transparency Report: YouTube Community Guidelines
Enforcement (visited Nov. 3, 2022), tinyurl.com/5cp5npev.
21
Twitter,
Platform
Manipulation
(July
28,
2022),
perma.cc/UQ8R-P5C7; Twitter, Rules Enforcement (July 28,
2022), perma.cc/Y9HD-UH63.
22
24
platforms and their users to be inundated by malicious content. Rather than fix anything, S.B. 7072’s
disclosure obligation will only further exacerbate the
problem, as it gives ill-intentioned actors insight into
new tools and will have a chilling effect on the innovation of improved methods to detect and counter
such content.
***
Together, S.B. 7072’s substantive prohibitions,
disclosure requirements, and other provisions are designed to force a sea change in how platforms operate,
against the will of both the platforms and the users
who have come to rely upon them, and all because
Florida does not like the content of the messages that
platforms are serving to their users. As cross-petitioners ably explain—and as the Eleventh Circuit held in
large part—the First Amendment does not tolerate
such governmental regulation and burdening of private speech. To confirm this result, and to ensure that
other States do not follow in Florida’s path, the Court
should grant both Florida’s petition in No. 22-277 and
the cross-petition here, and review the legality of S.B.
7072 in its entirety.
25
CONCLUSION
The Court should grant the conditional cross-petition.
Respectfully submitted.
PAUL W. HUGHES
Counsel of Record
SARAH P. HOGARTH
ANDREW A. LYONS-BERG
ALEX C. BOOTA
McDermott Will & Emery LLP
500 North Capitol Street NW
Washington, DC 20001
(202) 756-8000
phughes@mwe.com
Counsel for Amici Curiae
1a
APPENDIX: LIST OF AMICI
•
Chamber of Progress
•
Consumer Technology Association (CTA)
•
Engine Advocacy
•
Global Project Against Hate and Extremism
(GPAHE)
•
HONR Network
•
Information Technology & Innovation Foundation (ITIF)
•
Interactive Advertising Bureau (IAB)
•
IP Justice
•
LGBT Tech
•
Multicultural Media, Telecom and Internet
Council (MMTC)
•
Safeguarding American Values for Everyone
(SAVE)
•
The Software & Information Industry Association (SIIA)
•
Stop Child Predators (SCP)
•
TechNet
•
Washington Center for Technology Policy Inclusion (WashingTECH)
This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.