The words spoken in the Unit 14 video, scene by scene, as its captions show them, with what the animations show marked [On screen: ...] where they play. Each speaker is named where the speaker changes. The Unit 14 reading covers the same material.
The narration is spoken by synthetic voices made for this course, each shown on screen as a drawn robot. No person is speaking.
Unit 14 — Copyright and attribution
Narrator: Unit 17 is about disclosure: saying how a piece of work was made. Two more questions sit beside it, and they are constantly run together.
Three questions
Rights. What may lawfully be done with this: copied, published, sold, built on. Attribution. Who should be credited. And disclosure. How the work was made.
[On screen: A single question, am I allowed to use this, splits along three arms into rights, attribution and disclosure. Each then answers differently about the same artefact: rights says yes under the licence, attribution says not as your own, and disclosure says say a model made it. Each is answered by a different authority: the licence, the terms and the law, which is partly unsettled; this course; and honesty, with the form set by your institution. The three answers do not agree, which is the point: a yes on one is often taken for a yes on all three. They come apart the other way too: something can be credited and still an infringement.]
They come apart in both directions, which is the useful thing to know. Something can be entirely lawful to use and dishonest to present as your own. Something can be scrupulously credited and still an infringement.
What this unit will not tell you
Copyright in generated output is genuinely unsettled, and it is unsettled differently in different countries. Three questions are open. First, whether generated output has copyright at all, and if so, who owns it. Second, whether training on protected material without permission is lawful, and how far a rights holder's opt-out should count. Third, what obligations attach to output that resembles an existing work without copying it. This unit reports no court outcome on training. Where the law stands depends on the country and the date, and both are yours to check. That is not a way of avoiding the question. It is the answer. A page telling you that generated output is not copyrightable, or that training on public material is fair, would be making a confident claim where the evidence does not support one. It would not separate the settled parts from the contested ones. That is exactly what this course trains you to catch.
What is settled
Some of it is reasonably settled, and it covers most of what you need. Copyright protects a particular expression, not an idea, a fact or a style. Your inputs do not stop being covered by whatever covers them. A licence that binds you still binds you. If output reproduces a substantial part of a protected work, the ordinary rules about reproduction apply. A machine producing it is no exemption. And terms of service bind you by agreement. They govern your relationship with the provider, and can restrict what you do with output. They do not themselves show that copyright exists. But where rights do exist, terms may license them, which means give permission to use them, or assign them, which means transfer them. And unlike the open questions, you can go and read them.
Not only copyright
Copyright is not the only right involved. When generated output imitates a real person's likeness or voice, or uses a brand's name or logo, other questions arise. For a likeness in the UK, there is no single image right. Protection, if any, comes from several separate rules: passing off, trade marks, contract and copyright. Passing off means presenting something as if it were connected with another person or business. Elsewhere, a separate publicity or personality right may exist. And a trade mark is its own matter: a copyright licence need not cover it. The Apache licence, for one, grants no permission to use the licensor's trade marks.
Attribution is the part you can get right
Nothing about the unsettled questions stops you giving an honest account. Attribution is a practice, not a permission. Credit what a person made. If your work rests on someone's argument, data, figure or code, cite them. How you found it is irrelevant. Never cite a source you have not opened. That is Unit 9, and it is where these two units meet: attribution rests on the citation being real. If you read only a report of a work, say so, as the course's reference list does.
Checker: The course's claim checker was run on the course's own list of claims, and passed. Every source the list cites has an entry in the reference list, and every claim with no source is marked unverified. That shows each citation points to a real entry, not that anyone read the sources.
Two failures
Narrator: "Written with AI assistance", and no citations. That says how it was made, and nothing about whose ideas it rests on. A full bibliography, and no mention of assistance. That says whose ideas, and nothing about how the text was produced. You owe both. And they go in different places: citations where the claim is made, disclosure about the work as a whole. They are not worth the same to a reader, though. A quotation with its locator is a certificate: the reader can open the source and check your claim without trusting you. "Written with AI assistance" cannot be confirmed from the text. Disclose because it is honest, and where a rule requires it; put the effort into what can be checked.
Labels a tool attaches
As of 2026, some tools can attach a provenance label of their own. A content credential is a signed record of how a file was made. A watermark is a statistical mark in generated text.
[On screen: Three file cards side by side. File 1 carries a teal badge, 'signed record': a content credential. File 2 is generated text whose words carry a blue statistical mark, with a meter for watermark detection. File 3 has no label. A line notes that a label records where a file came from, not whether it is true. The badge lifts off File 1 and its status becomes 'credential stripped'. File 3's status becomes 'provider never marks it'. File 2's text is reworded, its blue mark fades and the detection meter drops: 'detection weakened'. Then the three documents become identical and each reads 'no label detected'. The closing lines: a missing label shows nothing about how the work was made; three different histories, one identical result.]
Both have limits. A label records where something came from, not whether it is true. A credential can be stripped. A text watermark marks only the output of providers that apply one, and rewriting or translation weakens its detection. So a missing label shows nothing about how the work was made. None of this replaces your own disclosure.
Do not name the system as an author
Questioner: Why not name the system as an author?
Narrator: Not because of a rule about who counts as an author. For a reason this course can defend. An author is someone answerable for the work. Naming a system as an author states that something can answer for the claims. And nothing in the system can. Record the tool the way you would record any other instrument. In the account of how the work was made. If your referencing style asks for an entry for a tool or a retained chat, follow it. That does not make the system an author, and it does not replace citing the underlying sources.
The awkward case
[On screen: Six numbered citations produced by an AI system. Four are marked as read and two as not read. Of the four read, two are marked as used. A panel headed 'Cite these two' says to cite them, at the claims they support. A panel headed 'Cite neither group' says to cite neither the two read but unused, which are not sources for anything written, nor the two never read, which are not yours to vouch for. A panel headed 'And the system itself' says the system is not an author, because a bibliography answers whose ideas; that if a referencing style asks for an entry, that records a tool; and that what the system did belongs in the disclosure.]
You read four of six papers, used two, rewrote everything in your own words.
Questioner: So which of the six do I cite?
Narrator: Cite the two. Do not cite the two you did not read, and do not cite the two you read and did not use. Disclose the assistance separately. And if the framing came from the summary, say so. "The structure of section two follows an outline produced by an AI system, which I then revised" is a true sentence, it costs you nothing, and it is exactly what a reader is entitled to know.
Next
The activity gives you six questions to sort by who can answer them. Some this course can answer. Some your institution or a contract can answer. Some do not have settled answers at all, and some sit partly in two places. Knowing which is which is most of the skill. Then come the two failures, your own worked case and the checkpoints.
Closing, 20 seconds, no speech. A candidate record on the left, a check against an independent reference in the middle, and a bracketed claim on the right fade in side by side. Below them: Continue with the course material, ai.qyber.dev/edu-ai-literacy-ada-verification/, Course playlist: youtube.com/playlist?list=PLPsBWTfgvakM, All links are also in the description, and Keep your record. Check the next claim.