Podcast legal basics: the five things that actually create liability

Ranked by how often each one bites

The theme tune is the most common infringement on the platform

A recorded song carries two copyrights. The composition — notes and words — belongs to the songwriter and their publisher. The sound recording — that performance of it — belongs to whoever paid for the session, usually a label. Using a record in a podcast needs both: a synchronisation licence from the publisher and a master use licence from the label.

Podcasters get caught because the licences they already hold look as though they should cover it. ASCAP, BMI, SESAC and GMR licence the public performance of a composition; SoundExchange collects for non-interactive streaming. An episode is downloaded and played on demand, so a copy is reproduced and distributed — different exclusive rights, cleared directly with the owners rather than through a performing rights organisation.

A theme tune is the worst version of this because it recurs: one unclear licence attaches to the whole back catalogue, so a takedown over intro music arrives as a demand about two hundred episodes rather than one. The fix is a written library licence naming on-demand podcast distribution and unlimited episodes. Read those two lines — "royalty free" is a pricing model, not a grant of rights. Music licensing for video covers the same two-copyright structure.

Who owns the recording of a conversation?

Recording an interview creates two things, and they do not belong to the same person by default.

• The recording. Whoever arranges and fixes the sound recording is normally its author — on a podcast, you.
• The words. The interviewer holds copyright in the questions and in the selection and arrangement of the material; the guest holds copyright in their own verbatim answers. Neither of you owns the whole conversation unless something says so.

That split is why "can you take my episode down" is awkward to refuse on principle. Above copyright sits a separate right — publicity, and in some states privacy — in the guest's name, voice and likeness. That one is state law and varies, so a release drafted for one state may not be enough in another.

A guest release does three jobs: it licenses the guest's contribution, it consents to your use of their name, voice and likeness including in promotion, and it confirms you may edit, excerpt and republish in any format. One paragraph in an email, agreed before you press record, is enough for most shows. The model release form covers the likeness half and the content licence agreement the contribution half; what a model release actually does explains why consent and licence differ.

Naming a person: where "in my opinion" stops helping

Defamation requires a false statement of fact about an identifiable person, published to someone else, causing reputational harm, with the fault the law requires. Public officials and public figures must show actual malice — knowledge of falsity, or reckless disregard for the truth. Private individuals face a lower bar set by their state, usually negligence.

The widespread belief is that prefacing an accusation with "in my opinion" converts it into protected opinion. It does not. In Milkovich v. Lorain Journal Co. (1990) the Supreme Court declined to recognise a separate constitutional privilege for opinion, and gave the example directly: "In my opinion Mayor Jones is a liar" is actionable, because it implies the speaker knows facts making it true.

The habit that follows: say what you know and how you know it. "The complaint filed last March alleges he did X" is a true statement about a document, and the document is the evidence for it. "He did X" is a claim about the world you have to stand behind. Where a guest makes the allegation rather than you, republishing it is still publication. Is it defamation sets out the elements.

Clips and fair use, honestly rather than optimistically

Fair use is a defence, not a permission. It is not something you hold before a dispute; it is something a court decides you had, after you were sued, weighing four factors under section 107: the purpose and character of the use, the nature of the work, the amount taken, and the market effect.

Two consequences follow. First, the argument turns on whether the clip is the subject of your commentary or merely the content of your episode. Playing thirty seconds of a speech to analyse what was said is a real argument; playing a song because it fits the mood is not. Second, a good fair use argument does not stop a takedown — distributors act on complaints, and the episode comes down while you explain. Relying on it for recurring material is expensive even when you are right.

The uses section 107 names first — criticism, comment, news reporting, scholarship — describe the territory where most podcasts sit comfortably. They do not extend it to production convenience.

Sponsorship: the disclosure has to be audible

The FTC Endorsement Guides, at 16 CFR Part 255, require a connection between an endorser and a seller which might materially affect the weight or credibility of the endorsement — and which the audience would not reasonably expect — to be disclosed clearly and conspicuously. Payment counts, so does free product, and so does a personal or business relationship.

"Clearly and conspicuously" means difficult to miss and easily understandable by ordinary consumers, and an audible disclosure must be delivered in a volume, speed and cadence sufficient for them to hear and understand it. That rules out the two habits podcasts actually have: burying the disclosure in the show notes, and reading it at the end at double speed.

One thing is genuinely not a problem: an advertisement the audience already recognises as one needs no separate label, because that connection is expected. The exposure sits in the personal endorsement — "I actually use this" — whose value comes from listeners believing it is a recommendation. Affiliate disclosure rules covers the written version; the sponsorship agreement template sets out who answers for the claims.

The dispute nobody plans for: the name and the feed

A co-hosted show is three assets: the trademark in the name, the copyright in the back catalogue, and the RSS feed with the hosting accounts behind it. They can end up in different hands, and usually nobody has decided which.

The asymmetry is the hosting account. Whoever can edit the feed can redirect it, and a redirect moves the whole subscriber base in a day. Ownership arguments settle in months; the feed settles in an afternoon. The credentials, not the intellectual property, are what the fight is really about.

Two people running a show for profit with nothing in writing may well be a general partnership by default in their state, governed by rules on ownership, decisions and dissolution that neither chose. Trademark rights in the United States arise from use in commerce rather than registration alone — but "used by whom" is exactly what a split makes contentious, and a registration standing in one host's name is discovered at the worst moment. Four lines settle it: who owns the name, who owns the recordings, who controls the feed, and what an exit pays. A short partnership agreement is the usual vehicle; partnership agreement essentials covers the rest.

The shape all five share

Every one of these is a permission assumed rather than asked for: the music, the guest, the clip, the audience's understanding of the sponsor read, the other host's expectations about the name. None is expensive to obtain before recording, and all are expensive afterwards. The recurring ones deserve the disproportionate attention, because a podcast compounds its mistakes — the same unlicensed track under a four-year-old intro is one error repeated two hundred times, in a catalogue that is public and trivially searchable.