What free speech isn’t allowed? — What the free expression clause excludes

What the free expression clause covers and what it does not

Brief statement of scope: free expression clause

The free expression clause broadly protects speech, but courts have long recognized a few narrow categories that fall outside that protection. For a practical primer that summarizes protected and unprotected categories, see the Brennan Center primer Brennan Center for Justice.

Historic categories of unprotected speech

Historically recognized categories include incitement, true threats, fighting words, obscenity, and defamation. These categories are narrow and defined by Supreme Court doctrine rather than by a single statutory list.

As of 2026, courts still rely on the leading Supreme Court precedents to determine when speech falls into these categories, and judges apply those precedents to new contexts such as social media and AI-generated material.

Incitement: the Brandenburg test and how courts apply it

The three elements: intent, imminence, likelihood

The Brandenburg test requires intent to incite, imminence of lawless action, and a likelihood that the speech will produce that action, as set out in the Supreme Court opinion Brandenburg v. Ohio, and see Oyez’s case summary.

Examples and borderline cases

Courts distinguish between abstract advocacy of ideas and directed calls for immediate unlawful conduct; a general political rant that praises violence is usually protected unless it meets the three Brandenburg elements.

Lower courts analyze whether the speaker sought to produce imminent lawless action, whether the context made such action likely, and whether direct, specific words were used to prompt immediate conduct.

True threats: when speech is a serious expression of intent to harm

Legal standard and mens rea questions

True threats are communications a reasonable person would interpret as a serious expression of intent to commit unlawful violence, a standard developed and refined in cases including Virginia v. Black.

Court analysis often asks whether the speaker possessed the requisite state of mind and whether the context supported a credible threat.

Context and audience factors

Judges consider the audience, surrounding circumstances, and whether the message carried specific, credible details that would make a violent act seem likely.

Online messages raise particular issues because anonymity, rapid sharing, and lack of tone can confuse observers about intent and seriousness.

Fighting words: the narrow, rarely applied category

Origin in Chaplinsky v. New Hampshire

The fighting words doctrine comes from Chaplinsky v. New Hampshire and covers personally abusive words likely to provoke immediate breach of the peace.

Modern limits and examples

Modern courts apply the category very narrowly. Many insulting or hateful remarks remain protected because they do not meet the original narrow test for provoking an immediate violent response.

Examples that fail as fighting words include broad insults on social media where the audience is diffuse and there is no immediate, targeted confrontation.

Obscenity and the Miller test

The three-part Miller factors

Obscenity is judged under the three-part Miller test: whether the average person, using community standards, would find the work appeals to prurient interest, whether it depicts sexual conduct in a plainly offensive way, and whether it lacks serious literary, artistic, political, or scientific value, as explained in Miller v. California.

How community standards and value inquiries work

Courts must assess local community standards, which can vary by jurisdiction, and then weigh whether the work has any serious value that would protect it from an obscenity ruling.

Online distribution complicates the analysis because content may be accessible across many communities with different standards, and courts have struggled with how to apply the community prong to widely distributed digital material.

Defamation: public figures, actual malice, and private-figure standards

New York Times Co. v. Sullivan and actual malice

When a plaintiff is a public official or public figure, the Supreme Court requires proof of actual malice before a defamation award will stand, a rule established in New York Times Co. v. Sullivan.

State law differences for private figures

Private figures generally face lower fault standards under state law; procedures and remedies therefore vary across jurisdictions and can affect how an alleged defamation claim is brought and evaluated.

Readers assessing reported defamation claims should check whether the subject is a public figure and whether the complaint alleges actual malice, because that determines the constitutional burden on proof.

How these tests are being applied to online and AI-generated speech

Platform moderation and private liability questions

Court decisions continue to rely on the traditional precedents while confronting questions about platform moderation and the legal distinction between private moderation and state action, as discussed in contemporary primers Brennan Center for Justice.

AI content, deepfakes, and evidentiary challenges

Because platforms are private actors, moderation decisions are typically not state action subject to the First Amendment, but those private policies raise practical and normative concerns about consistency and transparency.

How courts weigh evidence: decision criteria judges use in First Amendment cases

Context, intent, and the factual record

Judges focus on context, the speaker’s intent or state of mind, and the factual record when applying tests like Brandenburg or the true threats standard, with evidence on timing and audience often proving decisive Brandenburg v. Ohio.

Standards of review and appellate considerations

Appellate courts review both legal conclusions and factual findings; readers should note whether a trial court’s factual record supported its legal application, since appellate rulings often hinge on those details.

Where a case raises constitutional questions, the appellate path can affect how broadly a precedent is applied by later courts.

Common errors and misconceptions about unprotected speech

Confusing offensive content with unprotected categories

Offensive or hateful speech is frequently protected unless it fits a specific unprotected category such as incitement or a true threat.

Misunderstanding public figure defamation rules

Obscenity is not merely offensive sexual content; it requires the Miller test to be met before it is outside First Amendment protection, and public-figure defamation requires a showing of actual malice.

Both points mean that many emotionally charged or distasteful statements remain constitutionally protected in the United States.

Practical examples and short scenarios

Scenario: political rally and incitement

Imagine a speaker at a rally who urges listeners to go immediately to a nearby building and break windows to stop a vote. Courts would apply the Brandenburg elements to assess whether that call amounts to criminal incitement, focusing on intent, imminence, and likelihood Brandenburg v. Ohio.

Scenario: social post interpreted as a threat

An online message that says, without context, I will find you and harm you might be examined as a true threat; courts will look at whether a reasonable person would view the communication as a serious intent to harm and whether the speaker’s state of mind supports that interpretation Virginia v. Black.

Scenario: material tested under Miller

Material distributed online that some communities find offensive will lead courts to examine community standards and whether the work has any serious artistic or political value before labeling it obscene under Miller Miller v. California.

How readers can evaluate news, rulings, and platform decisions

Checking primary sources

When you read about a case or a moderation decision, start with the primary source: the court opinion or a reputable primer that summarizes controlling tests such as the Brennan Center guide Brennan Center for Justice or the Legal Information Institute’s Wex entry on the Brandenburg test Brandenburg test, or visit our news page.

Red flags in summaries or opinion pieces

Watch for oversimplified accounts that omit the controlling test or the factual record, claims that a ruling broadly changed First Amendment law, or pieces that fail to note whether the subject was a public figure.

Use a short checklist to verify coverage: what test did the court apply, what facts did the court find, who was the speaker, and was the action by a state actor or a private platform.

Reading a court opinion quickly: the four parts to check

Holding versus dicta

Check the holding first; the holding is the legal rule that controls later cases, while dicta are observations that do not bind future courts.

Facts, test applied, and remedy

Scan the opinion for the factual findings that the court relied on, the precise test the court applied, and the remedy or disposition to understand the case’s practical effect.

Look also for concurring or dissenting views, which can signal areas of disagreement that might produce future changes in the law.

Summary and next steps for readers who want to learn more

Concise recap of core tests

In short, the main unprotected categories correspond to the Brandenburg incitement test, the true threats doctrine, Chaplinsky’s fighting words, the Miller obscenity test, and the New York Times Co. v. Sullivan rule for public-figure defamation New York Times Co. v. Sullivan.

Trusted primary sources to consult

For primary documents, consult the Supreme Court opinions for Brandenburg, Miller, Chaplinsky, New York Times Co. v. Sullivan, and Virginia v. Black, and use reputable primers to see how courts apply those precedents to online speech and AI content; see also ACLU Ohio’s summary.

Readers should rely on the original opinions and authoritative primers rather than summaries when accuracy matters, or contact.