Copyright Challenges For AI-Generated Children’S Books That Retell FilIPino Myths.
📘 Copyright Challenges for AI‑Generated Children’s Books Retelling Filipino Myths
Retelling myths — whether Filipino folk tales like Ibong Adarna, Maria Makiling, Bathala, Legend of the Pineapple, etc. — raises unique copyright questions when an AI system generates a story. These include:
🔹 1. Authorship & Ownership of AI‑Generated Works
Core Questions
Who is the author when AI writes the book?
Can AI hold copyright?
If a human prompts the AI, does that person own the work?
Is the output derivative of existing works or original?
Legal Concepts
Copyright protects original expressions, not ideas.
AI is not a legal person — so it can’t own copyright.
The human who contributes creative direction may be considered the author, depending on how much human input exists.
But if an AI merely rearranges or copies expressions from existing myth retellings, the output may infringe others’ rights.
⚖️ Case Laws & Analogous Decisions
Because specific cases on AI‑generated content are still emerging, the best analysis comes from analogous rulings in the U.S., UK, Canada, and Australia involving:
Computer‑assisted creative works
Derivative works
Authorship disputes
AI‑related copyright suits
📌 Case 1: Naruto v. Slater (“Monkey Selfie”) (2018, USA)
Facts
A macaque took a photographer’s camera and snapped photos. A wildlife photographer later published the images.
Issue
Who owns the copyright — the monkey or the photographer?
Holding
The court held that non‑humans cannot be authors under copyright law.
Only a human author can hold copyright.
Implication for AI Books
Since AI is not human:
AI cannot hold copyright in its own output.
Only the human creator (prompt engineer, editor, publisher) can claim authorship, if sufficient human creativity is shown.
📌 Case 2: Thaler v. Perlmutter (2022, USA) – “System v. Brigadier General Thaler”
Facts
A patent dispute arose where AI created an invention.
Issue
Whether AI can be listed as an inventor.
Holding
The court found AI cannot be an inventor under U.S. intellectual property law.
Implication
Though a patent case (not copyright), it reinforces the principle that AI is not a legal creator. By extension, output from AI needs human creative contribution to qualify for copyright.
📌 Case 3: Naruto v. Slater – Second Suit on Copyright (2018, USA)
Facts
The same “Monkey Selfie” images were registered for copyright.
Issue
Whether such registration was valid.
Holding
Copyright could not be granted under the macaque’s name.
The U.S. Copyright Office maintained that only human authorship is valid.
Implication
Reinforces that outputs created by non‑humans (AI included) require human authorship to qualify for protection.
📌 Case 4: Fox v. Disney (2020, USA) – “Star Wars Beam Case”
Facts
An artist sued major film studios alleging characters were copied from his earlier creations.
Issue
Whether the later works were substantially similar and infringed his copyright.
Holding
The court applied the “substantial similarity” test.
Work was found to be derivative and infringing where core expressive elements were copied.
Implication
For AI‑generated Filipino myth books:
If the AI output reproduces copyrighted lines or unique narrative elements (plots, characterizations, details) from existing books, that may be infringement.
Myth ideas are not copyrighted, but expressions of those myths ARE (books, poems, retellings).
📌 Case 5: Authors Guild v. Google (2015, USA) – “Google Books”
Facts
Google digitized millions of books without express permission and made “snippets” available.
Issue
Was this fair use?
Holding
Fair use was upheld because:
The digital copies served a transformative, educational, and public interest purpose.
Implication
If an AI‑generated book uses copyrighted inputs in a way that transforms them (summarizing, re‑expressing in new form), it might be closer to fair use. However:
Use in a commercial children’s book is weaker as a defense than research access.
The nature of the transformation matters: merely paraphrasing may still infringe.
📌 Case 6: Designers Guild v. Russell Williams (2000, UK)
Facts
Textile designs were copied and sold with slight changes.
Issue
Were changes enough to avoid infringement?
Holding
No — minor changes still violated copyright.
Implication
AI retellings that only tweak wording but retain significant expression from existing books may be infringing.
⚖️ Key Legal Issues Specifically for AI‑Generated Filipino Myth Books
🟡 A. Derivative vs. Original Works
Copyright does not protect ideas or public domain myths, but it does protect expressions of them.
Myths like Maria Makiling are generally in the public domain if ancient or traditional.
BUT if the AI draws on modern copyrighted retellings (e.g., a book published in 2005 retelling the myth in a unique style), the output may be derivative and infringing.
Rule: If an AI’s output reproduces unique, original elements from another author’s copyrighted book—plots, dialogues, illustrations—it might be considered a derivative work requiring permission.
🟡 B. Substantial Similarity Tests
Courts often apply one of these:
Extrinsic Test – Looks at specific elements (plots, phrases, structure).
Intrinsic Test – Looks at the overall impression on the ordinary observer.
If an AI output is substantially similar in core expression—not just general theme—it may be infringing.
🟡 C. Fair Use / Fair Dealing
Factors include:
Purpose and character: Educational non‑profit use is stronger than commercial
Nature of work: Fiction is more strongly protected than fact
Amount taken: Small snippets are safer
Effect on the market: If the AI book replaces sales of the original, that weighs against fair use
For a commercial children’s book, fair use defenses are weaker.
🟡 D. Moral Rights
Even if an AI output is original enough to be protected:
Authors of original works used as inputs may still have moral rights:
Right of attribution – must be credited if significant expression was used
Right of integrity – work can’t be altered in ways that harm reputation
These rights are stronger in countries like Philippines, France, Canada, and the UK.
🧾 Practical Examples
Example 1: AI Retells Maria Makiling Using Public Domain Elements
Safe: If it creates its own narrative and doesn’t copy modern retellings.
Challenge: Must avoid mimicking unique expressions from recent books.
Example 2: AI Copies Dialogues from a Modern Book
Infringing: If it reproduces distinctive passages (even reworded) from a specific author’s book.
Example 3: AI Trained on Copyrighted Collections
If an AI training dataset included copyrighted retellings without permission:
Output may contain hidden copies
Raises concerns under global copyright laws
This is similar to the Authors Guild v. Google issues but focused on AI training rather than snippet publishing.
✅ Summary of Legal Risks
| 💥 Risk | 📌 Why It Matters |
|---|---|
| Derivative infringement | AI may reuse protected expressions without permission |
| No AI authorship | Only humans can hold copyright |
| Moral rights violations | Altering original works may harm author reputation |
| Fair use is weak for commercial children’s books | Favor educational research, not products |
| Dataset copyright issues | Training data legality matters |
🧠 Best Practices to Avoid Infringement
Train AI only on public domain or licensed materials
Use human oversight: edit and add creative direction
Avoid reproducing unique wording from copyrighted sources
Provide attribution when appropriate
Seek licenses for modern books used as sources
RELATED Blog
comments