“I can say whatever I want. It’s a free country.” That sentence contains one of the most common legal misunderstandings on Earth. Free speech is a right in almost every country, but it has never meant the freedom to say absolutely anything, absolutely anywhere, absolutely free of consequences. Every legal system on the planet draws lines. The interesting questions are where those lines sit, who decides what crosses them, and how much the answers change when you cross a border.
The short version: there is a global baseline that nearly every country has signed up to, a famously extreme protector (the United States), and a wide middle where the same words can be protected in one country and criminal in another. Here is how the limits actually work.
The global baseline: what the world’s treaty says
The closest thing to a world constitution on speech is Article 19 of the International Covenant on Civil and Political Rights, accepted by well over 170 countries. It has three paragraphs, and they do three different jobs [1].
Paragraph 1: your opinions can never be restricted. Not by any law, anywhere. You can think what you like, and the state cannot punish the thought itself. This is the absolute core.
Paragraph 2: your expression gets a broad right. Freedom to seek, receive, and share information and ideas of all kinds, in any medium, regardless of borders.
Paragraph 3: limits are allowed, but only through a three-part filter. A restriction on expression is lawful only if it ticks all three of these boxes:
- It is provided by law, meaning a clear written rule, not a vague custom or a minister’s mood
- It pursues a legitimate aim from a closed list: the rights and reputations of others, national security, public order, public health, or public morals
- It is necessary and proportionate for that aim, meaning no broader than the problem requires [1, 3]
Think of it like a speed limit. There must be a posted rule (legality). The rule must be about safety, not filling government coffers (legitimate aim). And a 5 km/h limit on an empty motorway would fail because it is unnecessary for its goal (necessity and proportionality). A restriction that fails any one box is a rights violation.
Article 20 adds a twist almost nobody expects: the treaty does not just permit limits, it requires some. States must prohibit propaganda for war, and any advocacy of national, racial, or religious hatred that constitutes incitement to discrimination, hostility, or violence. Silence in the face of that speech is itself a breach [1].
And the UN’s own guidance draws the bottom line just as firmly in the other direction. The UN human rights body has stated that expression regarded as deeply offensive is still protected, that general blasphemy laws are incompatible with international human rights law, that criticizing religious leaders cannot be prohibited, and that states may not impose a general ban on erroneous opinions or incorrect interpretations of past events [3].
The United States: the extreme protector
To understand limits globally, it helps to start with the country that has the fewest: the United States. American law protects categories of speech that most countries criminalize, and it does so through specific doctrines.
Incitement. In Brandenburg v. Ohio (1969), the Supreme Court set a two-part test that still governs: advocacy of illegal action can be punished only if it is (1) directed to inciting or producing imminent lawless action and (2) likely to actually produce it. Applied, the test has protected astonishing things: a Ku Klux Klan leader’s racist rally speech (conviction reversed, because it did not incite imminent action), and a protester’s shouted “we’ll take the street later” (protected, because “later” is not imminent). Even a civil rights leader’s heated boycott speeches, with threatening undertones, were protected [2].
Fighting words. The old Chaplinsky (1942) category of face-to-face insults likely to start a brawl still exists on paper, but later cases have narrowed it so severely that laws targeting it keep getting struck down as too vague or as discriminating by viewpoint [5].
The everyday result: in America, most hate speech, most offensive speech, and most angry political rhetoric are legal unless they are true threats or genuinely about to spark immediate violence. American neo-Nazis can march where German ones would be arrested. That is a real, deliberate choice, and it is the exception globally, not the rule.
How the rest of the world draws the lines
Most countries work from the Article 19 three-part test instead of the American categories. In practice that produces five big families of limits.
Defamation. Every system lets you be sued (and in many countries, criminally charged) for false statements that damage someone’s reputation. The difference is in the dials: many countries keep criminal defamation on the books, and protections for speech about public officials and public figures range from robust to almost none [1, 4].
Hate speech. Article 20 makes bans on incitement to discrimination, hostility, or violence mandatory. The UN’s six-part Rabat test helps courts decide where protected offensive speech ends and punishable incitement begins, weighing context, the speaker, intent, content, reach, and likelihood of harm. Below that top level, offensive but non-inciting speech must still be protected [3].
Threats and incitement to violence. Directly threatening a person is a crime everywhere, and inciting violence sits in the mandatory-ban zone of Article 20 [1, 3].
National security and public order. Every country restricts speech that endangers security or provokes disorder. The risk, watchdogs warn, is that these heads swallow everything else: the closed list of legitimate aims is broad, and states that interpret “public order” or “morals” aggressively can outlaw a great deal of criticism [1, 3].
“False news” and online content. This is the newest and fastest-growing limit. Singapore’s Protection from Online Falsehoods and Manipulation Act lets any government minister order correction notices or content removal for statements judged false or contrary to the public interest, backed by fines and up to a year in prison for non-compliance. It has been used against opposition figures, activists, and ordinary social media users. Singapore also allows defamation suits over general criticism of the government, which a 2019 ruling effectively confirmed, and has convicted critics of criminal defamation; in 2021 its media regulator cancelled the license of an independent news site, and another outlet’s staff were ordered to pay roughly US$156,000 in defamation damages [4].
The four ways limits are actually drawn
Strip the details away and every system limits speech in one of four ways.
1. Content bans. Whole categories (defamation, true threats, incitement, obscenity, fraud) sit outside protection. These are the bluntest tools.
2. Conduct-plus-speech. Speech can be limited when it merges into action: blocking a hospital entrance, burning a draft card as protest (the US casebook itself lists this example). The state regulates the conduct, and the speech rides along [5].
3. Speaker-based limits. Governments, soldiers, students, and employees accept restrictions that private citizens do not.
4. Platform and information controls. Rather than punishing speakers, the state orders platforms to correct, label, or remove content, as Singapore’s POFMA and FICA laws do. The speaker may never be charged, yet the speech still disappears [4].
An everyday guide: what this means for something you might actually say
- Criticizing the government: protected in principle everywhere, including erroneous opinions under the UN’s reading. But the practical risk varies enormously; in Singapore-style systems, criticism invites defamation suits and correction orders [3, 4]
- Organizing a boycott or a protest: American courts protect heated protest rhetoric; other systems regulate assemblies and may treat economic disruption as public-order offenses [2, 3]
- Racist or religiously insulting rants: legal in the US unless they incite imminent lawlessness; criminal in many countries if they meet the Article 20 incitement threshold; but merely offensive rants are protected even under international law [2, 3]
- Accusing your neighbor of a crime: defamation risk everywhere; truth is a defense almost everywhere, but the burden and the damages differ wildly [1]
- Threatening someone: a crime everywhere [1, 3]
- Sharing an unverified claim online: protected speech in the US; subject to government correction or removal orders in Singapore and an increasing number of countries adopting the same model [4]
Myths vs facts
| Claim | Verified answer |
|---|---|
| “Free speech means I can say anything.” | Myth. Every system limits defamation, threats, incitement, and more; the differences are in where the lines sit, not whether they exist [1]. |
| “My opinions can be made illegal.” | Myth. Opinions, as opposed to expression, cannot be restricted at all under Article 19(1); only the sharing of ideas can be limited, and only through the three-part test [1, 3]. |
| “Hate speech is banned everywhere.” | Myth. Only incitement to discrimination, hostility, or violence must be banned. Deeply offensive speech short of that threshold is protected under international law [3]. |
| “Hate speech is legal everywhere.” | Just as wrong. Article 20 makes bans on that incitement mandatory; many countries criminalize Holocaust denial, racial incitement, and similar speech that the US protects [1, 3]. |
| “Offensive speech can be outlawed.” | Myth under the global standard. The UN body has said expression regarded as deeply offensive remains protected, and general blasphemy laws are incompatible with international law [3]. |
| “American rules apply on the internet worldwide.” | Myth. The Brandenburg test is American; elsewhere the Article 19 necessity-and-proportionality test governs, producing very different outcomes for the same words [1, 2, 3]. |
| “Wrong opinions can be generally banned.” | Myth. The UN body states the covenant does not permit general prohibitions of erroneous opinions or incorrect interpretations of past events [3]. |
| “If speech is restricted, it must be because it is false.” | Myth. Truth is a defense in defamation almost everywhere, but much restricted speech (incitement, threats, hate speech meeting the Article 20 threshold) is restricted because of its danger or target, not its falsehood [1, 3]. |
Practical takeaways
- Never assume your country’s speech rules travel with you. A post legal at home can be a crime where your audience lives
- Truth and evidence matter more the further you get from the US model: in defamation and false-news systems, documentation is your shield
- The three-part test is your analytical tool: when you see a speech restriction, ask “What law? What legitimate aim? Is it necessary and proportionate?”
- Offense is not a crime under the global standard, but incitement is. Learn where the UN’s six-part Rabat test draws that line before assuming a rant is safe [3]
- Watch the platform layer: correction orders and content removal can silence speech even where the speaker is never prosecuted [4]
Bottom line
Free speech is best understood not as a wall but as a fence with a gate. What sits inside is enormous: opinions, criticism, satire, error, offense, and ideas most people dislike. What sits outside, in every country on Earth, is a family of harms: lies that destroy reputations, words that threaten, advocacy that pushes crowds toward violence, and propaganda that recruits for war. The countries of the world have signed the same treaty describing that fence, and the fiercest arguments are about where to place a few specific planks, most of them involving hate, religion, and the speed of the internet. The right is real, the limits are real, and knowing both is the actual meaning of the free-speech guarantee [1, 3].
Disclaimer: This article is for general education only and is not legal advice. Speech and media laws vary dramatically by country and change frequently; for any real situation involving publication, protest, or an investigation, consult a qualified lawyer in the relevant jurisdiction. If you notice an error or would like to send feedback or corrections, please contact us.
References
- [1] International Covenant on Civil and Political Rights, Articles 19 and 20, Office of the UN High Commissioner for Human Rights
- [2] Brandenburg test, Wex, Cornell Law School Legal Information Institute
- [3] Special considerations for hate speech, UNHCR
- [4] Singapore: Freedom in the World 2022, Freedom House
- [5] Fighting words: Chaplinsky v. New Hampshire and later narrowing, First Amendment casebook (H2O OpenCasebook)
