Lesson 21 of 1570
Copyright and AI: Who Owns What?
Generative AI trained on copyrighted work has triggered the biggest wave of copyright lawsuits in the internet era. Here is the state of the fight.
Lesson map
What this lesson covers
Learning path
The main moves in order
- 1Three Questions, No Settled Answers
- 2copyright
- 3fair use
- 4NYT v. OpenAI
Concept cluster
Terms to connect while reading
Section 1
Three Questions, No Settled Answers
Copyright law around AI is being written in real time by judges, legislatures, and lawyers who are all three steps behind the models. Three questions are at the center of almost every fight.
- 1Was it legal to train on this copyrighted data without permission?
- 2Does the AI's output infringe anyone's copyright?
- 3Who owns the output — the user, the company, or nobody?
Question 1: the training question
AI labs argue that training on copyrighted text is transformative fair use, the same doctrine that lets Google Books show snippets. Rights holders argue that AI is different: it produces substitutes for the original work, not snippets, and it was trained on billions of works without licenses.
Case: Getty v. Stability AI (UK, Nov 2025)
Getty Images sued Stability AI in the UK, alleging Stable Diffusion was trained on millions of Getty photos. On November 4, 2025, the English High Court handed down a split ruling. Copyright claims were largely rejected — the court held that model weights are not a copy of the training images. Getty won on a narrow trademark claim for some watermarked outputs. The ruling is the UK's first word on AI training and has already been cited in other jurisdictions.
Question 2: the output question
If an AI can generate something that looks a lot like a specific copyrighted character or style, is the output infringing? Courts are starting to say it depends. Generating a picture labeled Pikachu? Likely infringing. Generating a picture in an anime style that happens to resemble one artist? Unclear.
Question 3: the ownership question
In the US, purely AI-generated work cannot be copyrighted. The US Copyright Office has ruled that human authorship is required (2023 guidance, confirmed through 2025). If you prompt an AI and nothing else, the result is public domain. If a human creatively selects, arranges, or edits, the human contribution is copyrightable — but not the AI parts.
Compare: how the three regions are handling it
Compare the options
| Region | Training on copyrighted data | AI output copyrightability |
|---|---|---|
| US | Contested in court, fair use unresolved | No copyright without human authorship |
| EU | Text & data mining allowed, rights holders can opt out | AI Act requires disclosure of training content |
| UK | Contested (Getty 2025) | Generally requires human authorship |
| Japan | Broadly permits training for ML | Generally requires human authorship |
What artists and writers are doing
- Opting out via robots.txt and ai.txt where platforms honor it
- Using tools like Glaze and Nightshade that subtly alter images
- Licensing deals: Reddit, AP, News Corp, Shutterstock have signed
- Collective bargaining: WGA 2023 strike won AI-use protections
“We are not afraid of new technology. We are afraid of having our life's work taken without consent or compensation.”
Key terms in this lesson
The big idea: copyright and AI is an active legal battlefield, not a settled rulebook. What is allowed today may be illegal next year, and vice versa. If you build with AI, keep one eye on the courts.
End-of-lesson quiz
Check what stuck
15 questions · Score saves to your progress.
Tutor
Curious about “Copyright and AI: Who Owns What?”?
Ask anything about this lesson. I’ll answer using just what you’re reading — short, friendly, grounded.
Progress saved locally in this browser. Sign in to sync across devices.
Related lessons
Keep going
Builders · 25 min
AI and Homework: Where Is the Honest Line?
Using AI on schoolwork is not simply cheating or not cheating. It depends on the task, the rules, and what you are learning to do. Here is how to think about it.
Builders · 30 min
The EU AI Act in Plain English
The world's most ambitious AI law passed in 2024. Here is what it actually does, when it kicks in, and why it matters if you do not live in Europe.
Builders · 30 min
Catastrophic Risk, Without the Panic
Measured people at serious labs and universities publicly worry about AI going very wrong. Here is what they mean, what they disagree about, and how to read the headlines.