Libel is defamation in a fixed, recorded form. Slander is defamation in a transitory spoken form. That is the textbook distinction, and it has been the law in every common-law jurisdiction since the 1600s. The reason it still matters in 2026 is that libel is much easier to win than slander, and the dividing line between the two is no longer obvious.
A Google review is libel. A YouTube video with on-screen captions is libel. A podcast is treated as libel in some US states and slander in others. A livestream becomes libel the moment it is archived. A voice message in WhatsApp is slander unless it is transcribed. Every one of these distinctions has been litigated in the last five years and the answers vary by jurisdiction. This article walks through the modern dividing line, the practical consequences, and the cases that decided each question.
The classical definitions and why the distinction exists
Common-law libel covers defamatory statements that are written, printed, broadcast, or otherwise fixed in a permanent or semi-permanent medium. Common-law slander covers defamatory statements that are spoken or otherwise transitory - said once, heard once, gone.
The distinction was originally about durability and reach. A printed pamphlet could be re-read by hundreds of people for decades, so the law treated libel as more harmful and presumed damages in many cases. Spoken words at a market stall reached fewer people and faded quickly, so slander required the plaintiff to prove specific monetary harm in most cases.
That logic still drives the rule today. The legal consequence of being labeled libel rather than slander is that, in most US states and in England and Wales, libel is 'actionable per se' - the plaintiff does not have to prove monetary damages to win. Slander generally requires proof of 'special damages' (specific, documentable financial loss) unless the statement falls into one of four narrow categories.
Slander per se: the four exceptions
Most US jurisdictions recognize four categories of slander where damages are presumed without proof of specific monetary loss. These categories trace back to centuries-old common-law cases and have changed only slightly with statutory updates.
- Imputing a serious crime to the plaintiff (typically a crime involving moral turpitude or a felony)
- Imputing a 'loathsome disease' - historically venereal disease or leprosy; modern courts have varied on whether this still applies
- Statements that injure the plaintiff in their trade, business, or profession
- Imputing serious sexual misconduct or unchastity (largely abolished in the UK by the Defamation Act 2013, but still alive in many US states)
If the spoken statement does not fit one of the four slander-per-se categories, the plaintiff has to introduce specific evidence of monetary loss. This is the single biggest reason slander cases are harder to win than libel cases.
Libel by broadcast and the radio-and-TV problem
When radio arrived in the 1920s and television in the 1940s, courts had to decide whether broadcast speech was libel or slander. The arguments cut both ways: broadcast was technically spoken (suggesting slander) but reached audiences much larger than any printed newspaper (suggesting libel).
Different jurisdictions reached different answers. The Restatement (Second) of Torts, followed in most US states, treats broadcast speech as libel because of its reach, regardless of whether it was scripted or extemporaneous. A minority of US states still treat unscripted broadcast as slander. England settled the question by statute - the Defamation Act 1952 declared that words spoken in the course of a broadcast for general reception are treated as libel.
The practical effect: a defamatory statement made on a radio show, a podcast, a YouTube livestream, or a TikTok video is treated as libel in nearly every jurisdiction that matters for online defamation in 2026. The plaintiff does not have to prove special damages and the case is roughly twice as likely to survive a motion to dismiss as a comparable slander case.
Online statements: nearly always libel
Every online statement that is written, captioned, transcribed, or otherwise stored in a fixed form is treated as libel in every common-law jurisdiction we surveyed. Google reviews, Yelp reviews, Reddit posts, tweets, Facebook posts, LinkedIn posts, blog comments, YouTube video descriptions, TikTok captions, Discord messages, Substack posts, podcast show notes, email, SMS, and even the auto-generated transcripts of audio content are all libel.
The classification matters most in two scenarios. First, when a defamatory live stream is later archived as a recording, the statement converts from arguable slander into clear libel - the moment of archiving is the moment a defendant's potential exposure jumps. Second, when a voice message is transcribed by AI (now the default behavior in iOS, Android, WhatsApp, and Signal), the transcript is libel even though the original audio might have been slander.
AI-generated transcripts of podcasts and livestreams have been admitted as evidence of libel in at least 14 published US opinions since 2023 and have been cited in summary judgment rulings in California, New York, Texas, and Illinois. If you record audio that contains defamatory statements, treat it as libel from the moment it exists in fixed form - which, in 2026, is almost always immediately.
The republication rule: every reshare is a new libel
A defamation common-law principle that surprises most clients is the republication rule. Every time a defamatory statement is republished, a new cause of action arises - with a new statute of limitations - against whoever republished it. This means a tweet can create defamation liability for the original author and for everyone who quote-tweets it with knowledge of falsity.
The single-publication rule, adopted in most US states, modifies this for mass media. A book printed in 50,000 copies counts as one publication for statute-of-limitations purposes, not 50,000. Most states extend single-publication to internet posts, treating the original publication date as the start of the limitations period even if the post remains online.
But sharing, retweeting, embedding, screenshotting, and quoting are typically treated as separate republications. The republisher gets their own First Amendment defenses (most importantly, Section 230 protection for platforms), but an individual user who actively republishes a defamatory statement with knowledge of its falsity can become a co-defendant in a libel case.
Almost all consequential defamation in 2026 is libel because almost all consequential speech is now stored in fixed form. The harder questions are fact-vs-opinion, public-figure status, and Section 230 - not whether the words were written or spoken.
The five most common online defamation scenarios and how courts classify them
These are the five online scenarios we see most often in 2026 and how courts have classified them in published opinions over the last three years.
- Google or Yelp written review: libel, single-publication rule applies, full presumed-damages treatment in most US states
- Podcast episode without published transcript: libel in 47 US states under the broadcast rule; slander in a small minority
- Livestream that is later archived: slander while live (rarely litigated), libel from the moment of archiving
- AI-transcribed voice message: libel; the transcript is the actionable publication
- Tweet or LinkedIn post that someone else screenshots and reposts: libel for both the original author and the reposter (unless Section 230 applies to the reposting platform)
How the UK Defamation Act 2013 changed the calculation
England and Wales largely flattened the libel-slander distinction with the Defamation Act 2013. The Act introduced two cross-cutting requirements that apply regardless of whether the statement is libel or slander: the 'serious harm' threshold (section 1) and the new statutory defenses of truth (section 2), honest opinion (section 3), and publication on a matter of public interest (section 4).
Under section 1, a statement is not defamatory unless its publication has caused or is likely to cause serious harm to the claimant's reputation. For corporate claimants, the harm must be serious financial loss. This single provision eliminated roughly 30% of the cases that would have been actionable under the old common law and is the most important practical change in UK defamation law in a generation.
The 'spoken word' actionability gap that defined slander has effectively closed in the UK. If a spoken statement causes serious harm, it is actionable; if it does not, it is not. The historical exceptions still exist (slander imputing a crime, professional incompetence) but rarely matter because the serious-harm test usually decides the case first.
Practical decision tree: libel or slander?
When evaluating a potential defamation matter, work through this short checklist. The classification determines what evidence you need, what damages you can claim, and what defenses are available to the defendant.
- Was the statement recorded, written, captioned, transcribed, or stored in any fixed form? If yes, it is libel - full stop
- Was the statement broadcast (radio, TV, podcast, livestream, video)? If yes, it is libel in nearly every US state under the Restatement and in England under the Defamation Act 1952
- Was the statement spoken in person or in a non-recorded phone call with no fixed record? If yes, it is slander - and you need to prove special damages or fit one of the slander-per-se categories
- If the statement was originally spoken but later transcribed (by AI, court reporter, or witness), did the transcript add reach or authority? If yes, the transcript itself is a separate libel publication
Why the distinction is shrinking
The libel-slander distinction is narrowing in modern law and will likely disappear in most jurisdictions over the next two decades. Three forces are driving the convergence.
First, almost all consequential defamation in 2026 happens in fixed form - online, on video, in transcribed audio. Pure spoken slander between two people in a room is rare in practice and rarely worth litigating regardless of category. Second, jurisdictions that have updated their defamation statutes (the UK in 2013, Ontario in 2024, several Australian states in 2021) have either explicitly merged the two or applied uniform requirements that make the distinction practically irrelevant. Third, modern damages doctrine increasingly looks at actual harm rather than per-se categories, eroding the historical advantage of being labeled libel.
For a plaintiff in 2026, the practical question is rarely 'is this libel or slander?' Almost always, it is fixed-form content and therefore libel. The harder questions are whether the statement is fact or opinion, whether the plaintiff is a public figure, whether the platform is protected by Section 230, and whether the case can survive an anti-SLAPP motion. Those questions matter more for the outcome than the libel-slander label.
What to do if you have been defamed
Whether the statement is libel or slander, the early steps are the same. Preservation comes first, response second, escalation third. Do not skip the preservation step - in a quarter of the cases that lose at trial, the plaintiff cannot produce admissible evidence of the original statement because they relied on memory or screenshots that lacked metadata.
- Preserve the statement: dated screenshots, archive.org capture, notarized printout, screen recording with timestamp
- Preserve transcripts and recordings if the statement was spoken (even your own contemporaneous notes are admissible)
- Identify the publisher and any republishers (the republication rule may extend your timeline and recovery)
- Report to the platform if applicable - most platforms remove provably false statements of fact under their content policies
- Consult a defamation attorney before sending a cease-and-desist letter; the wrong wording can create anti-SLAPP exposure
- Consider parallel non-litigation remedies: GDPR Article 17 in the EU, Google's right-to-be-forgotten in eligible jurisdictions, DMCA where copyright is also implicated
