What online defamation is — and what it is not

Defamation is one of the oldest causes of action in the common law and one of the most consistently misunderstood. The misunderstanding is not usually about the technical rules. It is about what the tort is for.

Defamation protects your reputation in the estimation of other people. It does not protect you from embarrassment, from unfairness, from having your private life discussed, or from being written about in terms you find distressing. Those things may be actionable — under privacy law, data protection law, harassment law, or a platform’s own policies — but they are not defamation, and treating them as defamation tends to produce a weak claim, an unnecessary bill and a published record of the attempt.

This article sets out the boundary. It describes the elements a claimant must establish, the defences that decide most cases before they are ever filed, and the situations that feel defamatory but are governed by something else.

The three elements

Across common-law jurisdictions, including Kenya, a claimant must establish three things.

1. A defamatory meaning

The words must be capable of lowering the claimant in the estimation of right-thinking members of society generally, or of causing them to be shunned or avoided, or of exposing them to hatred, ridicule or contempt.

Two points do most of the work here.

Meaning is objective. The question is not what the publisher intended, and not what the claimant felt on reading it. It is what the ordinary, reasonable reader would understand the words to mean — a reader who is neither unusually suspicious nor unusually naive, who reads the publication as a whole rather than dwelling on one line, and who does not engage in strained or forced interpretation. A publisher who says “that is not what I meant” is generally not answering the right question. Nor is a claimant who says “everyone who knows me will realise what is being implied” — unless the implication is one the ordinary reader would draw.

Meaning can be indirect. Words innocent on their face may carry a defamatory meaning to readers who know additional facts. This is the “true” or “legal” innuendo, and it must be pleaded with the extrinsic facts and the readership who knew them. A photograph, a caption, an emoji, a juxtaposition, or a headline that a body of text later qualifies can all carry meaning. Kenya’s Defamation Act (Cap 36) defines “words” at section 2 to include “pictures, visual images, gestures and other methods of signifying meaning” — wide enough for memes, video and image posts.

Not every insult is defamatory. Vulgar abuse that no reasonable reader would take as an assertion of fact about the claimant’s character or conduct is generally not actionable, however unpleasant it is to receive.

2. Identification

The publication must be of and concerning the claimant. Naming is the obvious case, but it is not required. The test is whether reasonable readers who know the claimant would understand the words to refer to them. An unnamed “a senior partner at a Nairobi firm” may identify a person perfectly well if the surrounding detail narrows the field.

Group references are harder. The larger and more indeterminate the group, the less likely it is that any individual member is identified. A statement about “advocates” identifies no one; a statement about “the three directors of a named company” identifies each of them.

3. Publication

The words must be communicated to at least one person other than the claimant. Online, this element is rarely in issue — but it produces consequences that matter.

Every republication is a fresh publication. The original poster, the person who shares or quotes it, the platform that hosts it and, in principle, anyone who repeats the allegation may each be a publisher. That is not an invitation to sue everyone. It is a reason to think carefully about who is worth pursuing, who is merely a conduit, and which defendant actually controls the outcome you want.

A message sent only to the claimant is not published in the relevant sense. An abusive direct message with no third-party recipient may found other claims, but not defamation.

The defences that decide most cases

Most reputational disputes are resolved — or should be abandoned — on the defences, not the elements. Assess them honestly and early.

Truth (justification). If the substance of the imputation is true, the claim fails, however damaging the publication and however malicious the publisher. This is the single most important filter. The correct question is not “is every word accurate?” but “is the sting of the allegation substantially true?” Kenya’s Defamation Act section 14 preserves a justification defence in respect of multiple distinct charges where the unproven charges do not materially injure reputation given the truth of the rest.

Honest opinion (fair comment). Comment on a matter of public interest, based on facts that are stated or sufficiently indicated and are true, and which an honest person could hold, is protected — even if the opinion is exaggerated, obstinate or offensive. It is defeated by malice. Section 15 of the Defamation Act preserves the defence where the opinion is fair comment on the facts proved, even if not every alleged fact is proved. The frequent battleground is whether a statement is comment at all, or an assertion of fact dressed as an opinion.

Privilege. Some occasions attract absolute privilege — statements in court and parliamentary proceedings, for instance — where no claim lies whatever the motive. Others attract qualified privilege, where the publisher had a duty or interest in communicating and the recipient a corresponding interest in receiving, defeated by proof of malice. Fair and accurate reports of certain proceedings are separately protected by statute.

Apology, correction and offers of amends. Statutory schemes reduce or extinguish liability where a publisher moves quickly and properly. In Kenya, section 12 of the Defamation Act provides a defence for a newspaper libel published without malice or gross negligence where a full apology was inserted and a payment into court made when the defence was filed; section 13 provides an offer-of-amends route for unintentional defamation. These are demanding provisions with strict procedural conditions, but they mean a publisher’s early conduct materially changes the value of a claim.

What this is not

Several common situations feel like defamation and are not.

True but private. An accurate account of your medical history, relationships or finances is not defamation. It may be a breach of privacy under a constitutional right, an unlawful processing of personal data, or a breach of confidence.

Unflattering but not reputation-lowering. An article calling you unfashionable, boring or difficult to work with may sting without lowering you in the estimation of right-thinking people.

An image of you used without permission. This is a copyright, image-rights, privacy or data-protection question depending on who took the photograph and how it is used. Being in a photograph gives you no copyright in it.

A negative review of your product. Criticism of goods or services is often honest opinion. Where it crosses into a false factual assertion causing pecuniary loss it may be malicious falsehood or slander of goods, which have their own requirements and are not the same tort.

Impersonation and fake accounts. These are typically better addressed through platform impersonation policies and criminal law than through defamation, and far more quickly.

What online publication really changes

The doctrine is not different online. The practical picture is.

Permanence and searchability. A newspaper libel fades; a search result does not. This drives claimants towards removal and de-indexing rather than damages, and it means the value of an early, proportionate letter is often higher than the value of a judgment two years later.

Anonymity. A great many online publications come from accounts that cannot be attributed. Identifying an anonymous publisher is a separate and often harder problem than establishing that the publication is defamatory, and it may require criminal complaint machinery or disclosure applications against intermediaries.

Cross-border publication. Content published from one country and read in another engages more than one legal system, with materially different rules. Kenyan and Emirati law diverge sharply on defamation — see Defamation in the UAE — and a strategy that is sound in one may be counterproductive in the other.

Evidence that disappears. Posts are edited and deleted, accounts are suspended, and screenshots are challenged. Preservation is not an administrative afterthought; it frequently determines whether a claim is provable at all.

The clock. Limitation periods for defamation are typically much shorter than for other torts. In Kenya an action for libel or slander may not be brought after twelve months from the date the cause of action accrued, under section 4(2) of the Limitation of Actions Act (Cap 22) — against three years for tort generally.

Start from the outcome

Before choosing a route, decide what result would actually resolve the problem. The available outcomes are not interchangeable:

• Correction or right of reply — fastest, cheapest, and often the only thing that genuinely repairs the record.
• Removal at source — the only outcome that takes content off the internet, and it requires the publisher or the host to act.
• De-indexing — reduces discovery through a search engine; the page stays live.
• Damages — compensates, but slowly and publicly, and revives the story.
• A judgment — often the instrument required to make a platform act, rather than an end in itself.

A claim that is doctrinally strong but aimed at the wrong outcome will disappoint. A claim that is doctrinally weak but pursued loudly will make things worse. Both are avoidable with an accurate early assessment.