The Threshold of Censorship: Navigating the Intersection of Harmful Speech Regulation and Freedom of Expression in the Global Information Age

Full report

On this page

The demarcation line between the legitimate regulation of harmful speech and the imposition of authoritarian censorship is one of the most intractable dilemmas in modern information policy. As human communication has migrated almost entirely to digital spaces governed by private intermediaries and transnational architectures, the sovereign power to control information has been fundamentally disrupted—and aggressively reasserted. This report conducts an exhaustive jurisprudential, philosophical, and empirical analysis of global freedom of expression. It does not proceed from the assumption that all speech restrictions are inherently illegitimate, nor does it concede that all state-mandated regulations are justified. Rather, it evaluates how disparate political systems balance the preservation of public order, human dignity, and democratic integrity against the fundamental right to seek, receive, and impart information.

I. Comparative Legal and Philosophical Traditions

The philosophical underpinnings of free expression dictate how states approach the restriction of speech. These traditions range from libertarian models prioritizing a decentralized marketplace of ideas to paternalistic models that view the state as the supreme arbiter of truth, social harmony, and institutional integrity.

The United States: First Amendment Absolutism and the State Action Doctrine

The American constitutional tradition remains globally anomalous in its fierce protection of speech, deeply rooted in a libertarian distrust of state power. In the United States, the First Amendment prohibits the government from engaging in viewpoint or content-based discrimination unless the speech falls into a few narrowly defined historic exceptions, such as incitement, true threats, and obscenity. This robust protection is vividly illustrated by historical confrontations with deeply offensive speech and hostile audiences, such as the events in Skokie and Cicero, Illinois. In 1977, Frank Collin and the National Socialist Party of America (NSPA) sought to march in Skokie, a village with a substantial population of Holocaust survivors1. To prevent the neo-Nazi demonstration, Skokie enacted three ordinances: a comprehensive parade permit scheme requiring $350,000 in liability insurance, a ban on the dissemination of materials promoting hatred, and a prohibition on military-style uniforms at political demonstrations1. The federal courts struck down these ordinances as unconstitutional prior restraints on free expression3. The judiciary ruled that the government cannot prohibit speech merely because it is profoundly offensive, emotionally distressing, or repugnant to the audience; content-based prohibitions are presumptively suspect and trigger strict scrutiny3. Decades earlier, civil rights marches in Cicero, Illinois, tested similar boundaries regarding the "heckler's veto"—the notion that a hostile audience's potential for violence could justify silencing a speaker. The American jurisprudential consensus firmly rejects the heckler's veto, demanding that the state protect the speaker rather than suppress the speech to maintain order. A critical component of American speech jurisprudence is the State Action Doctrine, which dictates that constitutional free speech guarantees restrict only government entities, not private actors6. Under the public function exception, a private entity is only held to constitutional standards if it performs a function traditionally and exclusively reserved to the state, such as the company town in Marsh v. Alabama6. Despite arguments that dominant social media platforms constitute a "modern public square," the U.S. Supreme Court explicitly reinforced in Manhattan Community Access Corp. v. Halleck that private entities providing forums for speech retain their own First Amendment rights to exercise editorial discretion, immune from government mandates to host speech they find objectionable8.

Europe: Militant Democracy and the Margin of Appreciation

Contrasting with the American model, the European tradition—shaped by the cataclysms of the 20th century—embraces the concept of "militant democracy." This doctrine permits the state to aggressively restrict speech and political organization that threatens the democratic order itself12. The European Convention on Human Rights (ECHR) recognizes freedom of expression under Article 10 but explicitly allows restrictions that are "prescribed by law and are necessary in a democratic society" for the protection of health, morals, reputation, or national security. The European Court of Human Rights (ECtHR) articulated its defining standard in Handyside v. United Kingdom (1976), noting that freedom of expression extends to ideas that "offend, shock or disturb the State or any sector of the population"13. However, Handyside simultaneously established the "margin of appreciation," a doctrine granting domestic authorities significant latitude to determine whether a restriction is necessary, particularly concerning public morals and religious sensitivities14. This allows European nations to criminalize Holocaust denial and specific forms of hate speech without violating the ECHR.

International Human Rights Law: The Rabat Plan of Action

International human rights law attempts to harmonize these divergent traditions. Article 19 of the International Covenant on Civil and Political Rights (ICCPR) guarantees freedom of expression, while Article 20(2) mandates the prohibition of "any advocacy of national, racial or religious hatred that constitutes incitement to discrimination, hostility or violence"17. To operationalize the threshold between protected offensive speech and prohibited incitement, the United Nations Office of the High Commissioner for Human Rights (OHCHR) developed the Rabat Plan of Action17. The Rabat Plan establishes a rigorous six-part threshold test to ensure that criminal sanctions are applied exclusively in the most severe cases20.

Asia, Africa, Latin America, and the Middle East

Beyond the Global North, jurisprudence reveals unique adaptations of speech regulation, often heavily weighted toward state security, institutional protection, and social harmony over individual liberty. In Asia, Singapore prioritizes social cohesion and systemic trust through sweeping legislative mechanisms like the Protection from Online Falsehoods and Manipulation Act (POFMA) and the Foreign Interference (Countermeasures) Act (FICA)23. POFMA empowers government ministers to unilaterally determine what constitutes a "false statement of fact" and issue legally binding correction directions, takedown orders, or access blocks25. The burden is shifted to the citizen or independent journalist to appeal the minister's fiat in court, effectively chilling opposition discourse by installing the executive as the ultimate arbiter of truth25. In China, the 2017 Cybersecurity Law mandates data localization and grants the state immense power to force network operators to monitor and halt the transmission of prohibited content28. This is enforced via Article 105 of the Criminal Law, which outlaws "subversion of state power," a statute routinely used to silence human rights defenders and political critics28. In Africa, South Africa's approach is deeply informed by the historical trauma of Apartheid. Under the Promotion of Equality and Prevention of Unfair Discrimination Act (PEPUDA), speech is rigorously balanced against the constitutional imperative of equality. In the landmark Qwelane v South African Human Rights Commission case, the Constitutional Court refined the definition of hate speech, striking down overly broad provisions that merely targeted "hurtful" speech while upholding strict restrictions on speech that actively promotes hatred and harm, thereby aligning domestic law with ICCPR Article 20 standards33. In Latin America, Brazil currently serves as a profound case study in the risks of judicial "militant democracy." Following severe institutional stress and an attempted insurrection by supporters of former President Jair Bolsonaro, the Brazilian Supreme Federal Court (STF), led by Justice Alexandre de Moraes, vastly expanded its powers through the Inquérito das Fake News (Fake News Inquiry)36. Originally launched to protect the Court from defamation, the inquiry evolved into an unprecedented mechanism to suspend the social media accounts of politicians, freeze assets, and issue global takedown orders to foreign tech platforms—frequently under gag orders and without traditional adversarial due process37. In the Middle East, speech regulation is frequently executed through a combination of cybercrime laws, blasphemy statutes, and strict public order mandates. Regimes utilize highly elastic definitions of "insulting the state," "harming national unity," or "offending religious values" to criminalize political dissent and independent journalism. The lack of independent judicial review in many of these jurisdictions means that the strongest civil-liberties concerns regarding arbitrary detention and the suppression of legitimate criticism are routinely realized.

II. Categorical Analysis of Speech Restrictions

The legitimacy of speech regulation heavily depends on the specific category of content. By examining fifteen distinct categories of speech, one can observe the profound tension between compelling state interests and fundamental civil liberties.

Harm and Physical Safety

Incitement to violence represents one of the most universally recognized categories for legitimate restriction. The strongest argument for restriction is that speech which directly causes physical violence or severe discrimination irreparably damages societal order and human life. The state, holding a monopoly on violence, must intervene to prevent citizens from mobilizing mobs against vulnerable groups. However, the most potent civil-liberties concern is that broad definitions of "incitement" are routinely weaponized against political dissidents. If the requirement of "imminence" is removed, any radical critique of the state or demand for systemic revolution can be framed as incitement. The global gold standard for navigating this tension remains the Rabat Plan's six-part test, which requires a temporal and causal nexus between the utterance and the physical harm21. True threats similarly pit physical safety against expressive freedom. Restricting threats is justified because they inflict severe psychological trauma, restrict the freedom of the targeted individual, and serve as precursors to physical violence. Conversely, determining what constitutes a "threat" can be highly subjective, risking the criminalization of hyperbole, dark humor, or heated political rhetoric. The U.S. Supreme Court addressed this in Counterman v. Colorado (2023), vacating a stalking conviction based on an objective "reasonable person" standard. The Court ruled that the First Amendment requires the state to prove a subjective mental state of recklessness—showing that the speaker "consciously disregarded a substantial risk that his communications would be viewed as threatening violence"39. While Justice Sotomayor warned that an objective standard risks overcriminalizing speech, Justice Barrett dissented, arguing that true threats carry no social value and an objective standard is sufficient to protect victims42. Terrorism-related content presents an existential challenge for states. The argument for restriction rests on the reality that terrorist organizations utilize the internet for recruitment, radicalization, and operational coordination. Disseminating instructional material for violence poses a direct threat to national security. The corresponding civil-liberties concern is that anti-terrorism laws frequently utilize nebulous definitions. Governments can easily label domestic civil society organizations, environmental activists, or independent journalists as "terrorists" to legitimize the purging of their digital presence and their physical incarceration.

Reputation, Morals, and Vulnerable Populations

Defamation laws seek to balance the freedom of the press with individual dignity. Individuals have a fundamental right to the protection of their reputation against demonstrably false and damaging claims. However, criminal defamation laws are a favored tool of autocrats to silence investigative journalists. Furthermore, the threat of ruinous civil litigation forces immense self-censorship. To protect democratic accountability, courts in liberal democracies mandate a higher threshold for defaming public officials, such as the U.S. standard requiring "actual malice" (knowledge of falsity or reckless disregard for the truth)44. In systems lacking judicial independence, defamation laws effectively outlaw the exposure of state corruption. Obscenity regulation is driven by the state's interest in protecting public morality, particularly shielding minors from explicit content. The primary civil-liberties concern is that morality is subjective and historically transient. Expanding obscenity definitions often results in the suppression of LGBTQ+ identities, reproductive health information, and avant-garde art. The ECtHR’s Handyside margin of appreciation acknowledges that moral requirements vary, yet this variance permits highly localized, majoritarian censorship13. Child Exploitation Material (CSAM) is a category where the argument for restriction is absolute. The production and distribution of CSAM inherently involve the physical and psychological abuse of children, yielding a universal consensus on its prohibition. The civil-liberties concerns do not revolve around the right to distribute such material, but rather the mechanisms used to enforce the prohibition. Mandating client-side scanning or breaking end-to-end encryption to detect CSAM poses catastrophic, systemic risks to the privacy, security, and free expression of all internet users. Hate speech silences marginalized groups, normalizes bigotry, and lays the psychological groundwork for atrocities. Human rights frameworks explicitly require its prohibition when it crosses into incitement21. However, "hate" is an emotion, not an objective legal standard. Banning hate speech often devolves into protecting majoritarian religious or political sensitivities, effectively acting as blasphemy laws by another name. The UN Strategy on Hate Speech explicitly notes that not all offensive communication crosses the threshold of illegality; much remains protected even if deeply offensive22.

State Security and Institutional Integrity

Wartime information is heavily restricted under the argument that during armed conflict, the state's survival is at stake. Controlling troop movements, casualty reports, and strategic data is a matter of life and death. The overriding civil-liberties concern is that wartime censorship is inevitably expanded to shield military incompetence, hide war crimes, and silence anti-war opposition, destroying democratic oversight precisely when it is most needed. Classified information laws exist because the unauthorized disclosure of state secrets can compromise intelligence networks, diplomatic relations, and national security infrastructure. Yet, over-classification is endemic across global governments. Without whistleblowers leaking classified information to the press, the public remains ignorant of state crimes, mass surveillance, and illegal wars. Strict espionage laws routinely fail to provide a public-interest defense for journalists and whistleblowers. Political propaganda and extremist content are restricted to prevent state-sponsored foreign disinformation operations and domestic radical networks from artificially manipulating public sentiment and destabilizing institutions. The danger to civil liberties lies in the inherent relativity of "extremism." If a state defines its current regime as the baseline of moderation, any radical demand for systemic change is categorized as extremist. Banning propaganda often results in the creation of a closed, state-managed epistemic bubble where only government-approved narratives can circulate.

Epistemic Integrity and The Digital Sphere

Misinformation and disinformation regulations are built on the premise that the viral spread of false information fractures shared reality and causes measurable real-world harm. Algorithms designed for engagement amplify falsehoods, leading many to argue that the traditional "marketplace of ideas" remedy—countering bad speech with more speech—is structurally obsolete. The civil-liberties concern is that giving the government the power to dictate objective truth is historically disastrous. Misinformation laws are routinely used to censor factually accurate information that contradicts official government narratives. Election speech is uniquely vulnerable. Coordinated campaigns to deceive voters about the time, place, and manner of voting, or to baselessly undermine the integrity of electronic voting machines, directly sabotage the democratic process37. However, restricting speech surrounding elections can inadvertently outlaw legitimate political debate, hyperbole, and skepticism of electoral processes. Democracy requires citizens to have the right to question the integrity of their institutions without facing criminal charges. Medical misinformation regulation aims to protect public health. During a pandemic, the proliferation of false cures or anti-vaccine conspiracy theories directly correlates with excess mortality. Singapore utilized POFMA extensively to issue correction notices regarding COVID-19 falsehoods, arguably saving lives by disrupting virality26. Conversely, scientific consensus is evolutionary, not static. Early in the COVID-19 pandemic, hypotheses that were heavily censored by platforms at the behest of governments were later deemed scientifically plausible. Government regulation of medical speech ossifies fluid scientific inquiry into state dogma. AI-Generated Material, including deepfakes, presents a novel frontier. Generative AI allows for the frictionless creation of photorealistic non-consensual intimate imagery and voice-cloned political deepfakes, which can permanently destroy reputations and swing elections in hours. The civil-liberties concern is that draconian AI regulations may criminalize parody, satire, and legitimate synthetic art. Furthermore, establishing the mens rea (intent) required for traditional speech crimes is exceedingly difficult when a probabilistic algorithmic system generates the final output.

III. The Exploitation of Broad Categories for Authoritarian Censorship

Authoritarian and hybrid regimes rarely ban "human rights" or "democracy" outright. Instead, they weaponize the very categories of speech regulation recognized as legitimate by the international community—national security, public order, and misinformation—to execute systemic censorship.

Russia: The Ultimate Case Study in Security Exploitation

The Russian Federation exemplifies the catastrophic weaponization of speech regulation. Following its full-scale invasion of Ukraine in 2022, Russia rapidly enacted sweeping war censorship laws. Article 207.3 of the Criminal Code criminalizes the public dissemination of "knowingly false information" regarding the use of the Armed Forces of the Russian Federation46. In practice, this law legally equates any deviation from the Ministry of Defense's narrative with "false information." Utilizing the terminology of "war" rather than "special military operation," referencing civilian casualties in Ukraine, or engaging in peaceful anti-war protests triggers draconian prison sentences46. By framing independent journalism as a national security threat and foreign propaganda, the Russian state has weaponized the concepts of "misinformation" and "wartime information" to completely eradicate domestic civil society and political opposition.

Systemic Exploitation in China, Singapore, and Brazil

China operates a highly sophisticated apparatus of total information control. The 2017 Cybersecurity Law mandates data localization and grants the state immense power to force network operators to monitor, censor, and halt the transmission of prohibited content28. This digital infrastructure is legally backed by Article 105 of the Criminal Law, which outlaws "subversion of state power"28. Any critique of the Chinese Communist Party, human rights advocacy, or mobilization of minority groups is swiftly categorized as state subversion, effectively blending public order regulations with absolute political censorship. Regimes need not be fully authoritarian to exploit broad categories. In Singapore, POFMA has been invoked overwhelmingly against opposition politicians, alternative media, and civil society25. While POFMA is framed as a shield against digital disinformation, its architecture allows any minister to target speech during an election cycle, forcing independent journalists to carry government-mandated correction notices that publicly discredit their reporting23. The lack of independent judicial involvement prior to the issuance of a POFMA order flips the presumption of free speech, creating a structural chilling effect across the media ecosystem25. Conversely, Brazil demonstrates the risks of judicial overreach in the name of democratic preservation. The STF’s Inquérito das Fake News bypassed traditional prosecutorial authorities, allowing a single Supreme Court Justice to act as investigator, prosecutor, and judge37. To combat right-wing disinformation and threats against the democratic order, the STF ordered platforms to block accounts entirely, rather than targeting specific unlawful posts37. The conflict escalated into extraterritorial demands, where the STF allegedly attempted to force U.S.-based companies to remove content hosted globally, invoking deep concerns about whether an aggressive defense of democracy is inadvertently adopting the mechanics of authoritarian censorship38. When public order and protection from harmful content are used to bypass due process, legitimate regulation devolves into censorship.

IV. Legal Tests and a Principled Framework for Adjudication

Distinguishing legitimate, narrowly tailored restrictions from authoritarian censorship requires rigorous adherence to a procedural and substantive framework. Without such a framework, speech regulation defaults to the subjective preferences of the ruling class. The principle of necessity and proportionality dictates that any restriction on speech must be absolutely necessary to achieve a legitimate aim and must be the least restrictive means available. A government cannot impose a total platform ban or an internet blackout if blocking a specific inciting channel would suffice37. Proportionality requires that the societal benefit of restricting the speech demonstrably outweighs the injury to the fundamental right of expression. The establishment of **intent (mens rea) and harm** is critical. As demonstrated in the U.S. Counterman decision and the UN Rabat Plan, the speaker's mental state is paramount. Accidental misinformation, negligent hyperbole, or offensive comedy lack the requisite intent to cause harm. A baseline of subjective recklessness—at a minimum—must be established to separate malicious actors from careless citizens21. Furthermore, the harm must be tangible and strictly defined; emotional distress or "hurt feelings" cannot form the basis of a criminal restriction without opening the door to the heckler's veto, as historically contested in cases like Collin v. Smith3. To prevent the state from criminalizing abstract ideological debates, the legal test of imminence must be applied. The causal chain between the speech and the unlawful act must be direct and immediate21. If there is time for counter-speech to diffuse the danger, the restriction is generally illegitimate. Finally, the greatest safeguard against censorship is procedural: due process, transparency, and independent judicial review. Executive fiat, such as Singapore's POFMA orders or automated algorithmic takedowns mandated by state regulators, lacks adversarial due process. Speech restrictions must be transparently enacted, publicly justifiable, and subject to immediate, independent judicial review. The adjudicating body must be entirely insulated from political pressure to prevent the weaponization of the law.

V. Difficult Edge Cases and Scholarly Disagreements

The boundaries of this principled framework remain fiercely debated among legal scholars and philosophers, particularly as technology reshapes the public sphere. A primary scholarly disagreement surrounds whether the platform should be treated as the state. As social media platforms become the primary arenas for democratic discourse, some argue they act as modern public utilities or state equivalents and should be bound by First Amendment obligations7. This camp argues that corporate censorship is virtually indistinguishable from state censorship in its effect on the public discourse. Others fiercely push back, arguing that compelling private companies to host neo-Nazi propaganda or medical disinformation violates the platforms' own editorial rights and endangers user safety, reaffirming the State Action Doctrine10. Another profound disagreement exists between the proponents of militant democracy and those of viewpoint neutrality. European scholars often argue that the U.S. model of viewpoint neutrality is dangerously naive in the face of coordinated fascist and authoritarian movements that use the rights of liberal democracy to destroy it from within12. They argue that democracies have a right to preemptively ban extremist political parties. Conversely, First Amendment scholars argue that giving the state the power to preemptively crush "anti-democratic" movements inevitably allows incumbents to label their political opponents as existential threats, resulting in a self-fulfilling destruction of democratic norms. Deplatforming global leaders presents an ongoing edge case. When private tech companies banned sitting political leaders (such as Donald Trump or Jair Bolsonaro) following instances of civic unrest, it raised unprecedented questions about jurisdiction, sovereignty, and the power of private actors to regulate political speech that elected judiciaries were hesitant to touch.

VI. Universal Safeguards and Conclusion

As the digital ecosystem evolves to encompass generative AI, hyper-realistic deepfakes, and algorithmic amplification, the speed and scale of information flow will tempt governments of all political structures to expand their regulatory reach. When does legitimate regulation become censorship? It crosses the threshold when the state shifts its focus from preventing demonstrable, imminent, physical harm to controlling the epistemic baseline of its populace and shielding its institutions from critique. To navigate this volatile era, the international community must adopt a baseline of proposed universal safeguards that can apply across varied political systems and the emerging landscape of AI regulation:

1. The Abolition of Executive Fiat in Speech Regulation: No political entity, intelligence agency, or executive minister should possess the unilateral authority to declare information "false" or "illegal" with immediate binding effect. All content removal orders must originate from, or be immediately subject to, an independent, adversarial judicial process.

2. The Prohibition of Broad Definitional Frameworks: Laws targeting "fake news," "subversion," or "extremism" without strict, granular definitions tied to actual imminent violence or demonstrable fraud must be recognized as structurally incompatible with international human rights law.

3. The Preservation of Subjective Intent in the AI Era: As AI generates vast quantities of synthetic speech, liability must remain firmly tied to human mens rea. Individuals should not be criminalized for the emergent, hallucinatory outputs of algorithmic systems unless subjective intent or extreme recklessness in the deployment and dissemination of that system is proven beyond a reasonable doubt.

4. The End of Extraterritorial Jurisdictional Overreach: States must be restrained from issuing global takedown orders that force private platforms to censor content beyond the restricting state's borders, preserving the sovereign right of diverse nations to maintain their own speech standards.

The defense of free expression does not require a naive blindness to the profound harms that weaponized speech can inflict on individuals, minorities, and democratic institutions. However, it demands a profound humility from the state. The arc of history demonstrates that governments, regardless of their democratic mandates, are inherently flawed arbiters of truth. Maintaining the delicate threshold between regulation and censorship requires an unwavering commitment to the principle that the cure for dangerous ideas is rarely the brute force of the state, but rather the rigorous, unfettered resilience of a free society.

Works cited

1. National Socialist Party of America v. Village of Skokie \- Wikipedia, https://en.wikipedia.org/wiki/National\_Socialist\_Party\_of\_America\_v.\_Village\_of\_Skokie

2. Nazis in Skokie: Fighting Words or Heckler's Veto?, https://via.library.depaul.edu/cgi/viewcontent.cgi?article=2460\&context=law-review\&httpsredir=1\&referer=

3. Collin v. Smith – Case Brief Summary \- Studicata, https://www.studicata.com/case-briefs/case/collin-v-smith

4. Collin v. Smith, 447 F. Supp. 676 (N.D. Ill. 1978\) \- Justia Law, https://law.justia.com/cases/federal/district-courts/FSupp/447/676/1621085/

5. Collin v. Smith, 578 F.2d 1197 (1978) \- UMKC School of Law, http://law2.umkc.edu/faculty/projects/ftrials/conlaw/collinvsmith.htm

6. Free Speech and Public Space After Occupy Wall Street, https://northwesternlawreview.org/episodes-vol-106/free-speech-and-public-space-after-occupy-wall-street/

7. The State Action Doctrine at the Nexus of First Amendment Rights, https://www.trojanreview.org/state-action-doctrine

8. Social Media, The Modern Public Forum: The State Action Doctrine, https://mhlawreview.org/article/social-media-the-modern-public-forum-the-state-action-doctrine-and-resurrection-of-marsh/

9. De Facto State: Social Media Networks and the First Amendment, https://scholarship.law.nd.edu/cgi/viewcontent.cgi?article=4889\&context=ndlr

10. Knowledge and Decisions in the Information Age, https://laweconcenter.org/resources/knowledge-and-decisions-in-the-information-age-the-law-economics-of-regulating-misinformation-on-social-media-platforms/

11. The First Amendment Limits State Actors—Not Private Companies, https://www.ocbar.org/?pg=OCL-Featured-Articles\&blAction=showEntry\&blogEntry=140691

12. Militant Democracy Unmoored? The Limits of Constitutional Analogy, https://academic.oup.com/ejil/article/35/2/411/7679437

13. Handyside v United Kingdom \- Wikipedia, https://en.wikipedia.org/wiki/Handyside\_v\_United\_Kingdom

14. Handyside vs United Kingdom | LawTeacher.net, https://www.lawteacher.net/free-law-essays/human-rights/handyside-vs-uk.php

15. View of Fifty years after Handyside \- Coventry University, https://publications.coventry.ac.uk/index.php/clj/article/view/849/909

16. The Margin of Appreciation in European Human Rights Law, https://www.diplomacyandlaw.com/post/the-margin-of-appreciation-in-european-human-rights-law

17. What is Hate Speech? | Rights for Peace, https://www.rightsforpeace.org/hate-speech

18. the role of Rabat and the importance of civil society space, https://www.universal-rights.org/implementing-resolution-1618-the-role-of-rabat-and-the-importance-of-civil-society-space/

19. Toolkit for analysing a case of hate speech, https://rm.coe.int/advanced-guide-toolkit-how-to-analyse-hate-speech/1680a217cd

20. ARTICLE 19 welcomes the Rabat Plan of Action on Prohibition of, https://www.article19.org/resources/article-19-welcomes-rabat-plan-action-prohibition-incitement-calls-full-implementation/

21. Election Obligations & Standards Database, https://eos.cartercenter.org/quotes/9679

22. The hate speech barometer \- FORB Learning Platform, https://www.forb-learning.org/exercises/the-hate-speech-barometer/

23. Submission to the Universal Periodic Review of Singapore, https://www.hrw.org/news/2025/10/09/submission-to-the-universal-periodic-review-of-singapore

24. Singapore: Freedom on the Net 2023 Country Report, https://freedomhouse.org/country/singapore/freedom-net/2023

25. How Singapore's "fake news" law gets exported \- The Ballot, https://www.theballot.world/articles/singaporefakenews

26. Has POFMA been effective? A look at the fake news law, 1 year, https://www.channelnewsasia.com/singapore/singapore-pofma-fake-news-law-1-year-kicked-in-688816

27. Civil Society Responses to Singapore's Online “Fake News” Law, https://ijoc.org/index.php/ijoc/article/download/17910/3593

28. Hate Speech And Criminal Law In China \- Law Gratis, https://www.lawgratis.com/blog-detail/hate-speech-and-criminal-law-in-china

29. China: Freedom on the Net 2019 Country Report, https://freedomhouse.org/country/china/freedom-net/2019

30. Human Rights in China and U.S. Policy: Issues for the 116th Congress, https://www.everycrsreport.com/reports/R45956.html

31. Country policy and information note: opposition to the state, China, https://www.gov.uk/government/publications/china-country-policy-and-information-notes/country-policy-and-information-note-opposition-to-the-state-china-january-2026-accessible

32. All Comments \- NORMLEX \- International Labour Organization, https://normlex.ilo.org/dyn/nrmlx\_en/f?p=1000:13101:0::NO:13101:P13101\_COMMENT\_ID:4467145

33. Qwelane v South African Human Rights Commission \- Wikipedia, https://en.wikipedia.org/wiki/Qwelane\_v\_South\_African\_Human\_Rights\_Commission

34. Hate Speech in the Equality Act Following the Constitutional Court, https://perjournal.co.za/article/view/15438

35. HCWG dismayed at Qwelane hate speech judgment by Supreme, https://hcwg.org.za/hcwg-dismayed-at-qwelane-hate-speech-judgment-by-supreme-court-of-appeal/

36. Brazil – The state of democracy in the Americas \- International IDEA, https://www.idea.int/gsod/2023/chapters/americas/case/brazil

37. How Courts Became a Battlefront Against Disinformation, https://verfassungsblog.de/how-courts-became-a-battlefront-against-disinformation/

38. Brazil Censors · Case No. 8:25-cv-00411, https://brazilcensors.com/

39. Assessing Online True Threats and Their Impacts, https://jolt.law.harvard.edu/digest/assessing-online-true-threats-and-their-impacts

40. True Threats and Protected Speech in Counterman v. Colorado, https://scholarship.law.uc.edu/cgi/viewcontent.cgi?article=1601\&context=uclr

41. Counterman v. Colorado | 600 U.S. \_\_\_ (2023) \- Justia Supreme Court, https://supreme.justia.com/cases/federal/us/600/22-138/

42. Counterman v. Colorado \- Harvard Law Review, https://harvardlawreview.org/print/vol-137/counterman-v-colorado/

43. Counterman v. Colorado (2023) | The First Amendment Encyclopedia, https://firstamendment.mtsu.edu/article/counterman-v-colorado/

44. In Counterman v. Colorado, the Supreme Court Confirms the Vitality, https://www.dwt.com/insights/2023/07/supreme-court-counterman-v-colorado-free-speech

45. Special considerations for hate speech \- UNHCR, https://www.unhcr.org/handbooks/informationintegrity/understanding-challenge/special-considerations-hate-speech

46. Russia's war censorship laws must go \- Amnesty International, https://www.amnesty.org/en/petition/russias-war-censorship-laws-must-go/

47. The Big Chill? How Journalists and Sources Perceive and Respond, https://www.tandfonline.com/doi/full/10.1080/1461670X.2023.2192299

48. the absurdities of the fake news inquiry / os absurdos do \- UNIFACS, https://revistas.unifacs.br/index.php/redu/article/viewFile/8643/5036