[MEGATHREAD] Using AI for Fiction Worldbuilding — Who Owns the Lore, Maps & Calendar Systems? (2026)

Over the past year we have seen a surge of questions from novelists, game designers, and indie worldbuilders about using AI tools — particularly Claude and ChatGPT — to help design the foundational elements of their fictional worlds. Calendar systems, magic systems, planetary physics, constructed languages, maps, and more. The IP ownership questions around these uses are genuinely different from the standard "who owns AI-generated text" discussion, so this thread consolidates the best advice in one place.

What makes worldbuilding different:

• Writers are often using AI as a research and calculation tool rather than a prose generator — computing orbital mechanics, checking linguistic consistency, or modeling tidal patterns
• The creative vision (the world concept, its cultures, its history) typically originates entirely with the human author
• AI outputs are integrated into a larger creative work (a novel, a game, a series bible) where the human authorship is substantial
• The line between "AI as calculator" and "AI as co-author" is blurry and legally significant

This megathread covers:

• Copyright vs. contractual ownership of AI-generated worldbuilding elements
• Specific scenarios: calendar systems, constructed languages, maps, magic systems, game mechanics
• The compilation doctrine and how it applies to AI-assisted fiction
• Publisher disclosure requirements for AI-assisted manuscripts
• Practical documentation strategies for protecting your IP position

Related threads:

• MEGATHREAD: AI-Generated Content, Copyright & Commercial Use
• AI-generated content copyright ownership
• Claude AI terms for business use
• AI scraping my content — legal options

Please keep discussion focused on the legal and practical dimensions of AI-assisted worldbuilding. This is not the thread for debating whether AI should be used in creative writing. New developments will be added as they arise.

Legal Framework: Ownership of AI-Generated Worldbuilding Elements

Before we get into specific scenarios, it is important to understand that there are two distinct questions people conflate when they ask "do I own this?"

1. Contractual Ownership vs. Copyright Ownership

Anthropic's terms of service (and OpenAI's, for that matter) assign you rights to the outputs generated by their models. Under the contract, you have the right to use, modify, and commercialize those outputs. This is a contractual right. It means Anthropic will not come after you for using what Claude produced. But contractual ownership is not the same thing as copyright ownership. A contract cannot create a copyright where the law says none exists.

2. The Human Authorship Requirement

Copyright requires "human authorship." The Copyright Office's Part 2 report on AI and copyrightability, finalized in January 2025, reaffirmed this position and provided additional guidance on the spectrum of human involvement. The core principle: the more the human directs, selects, arranges, and creatively shapes the output, the stronger the copyright claim. A bare prompt — "design me a 600-day calendar for a planet with 4 moons" — almost certainly does not give you copyright in the resulting calendar system. The AI, not you, made the expressive choices.

3. The Compilation Doctrine (Zarya of the Dawn)

This is where things get interesting for worldbuilders. In the Zarya of the Dawn registration decision, the Copyright Office held that while the individual AI-generated images were not copyrightable, the author's selection and arrangement of those images into a cohesive graphic novel was copyrightable as a compilation. Applied to worldbuilding: even if individual AI-generated elements (a raw calendar, a list of vocabulary words, a set of tidal calculations) are not independently copyrightable, your creative integration of those elements into a novel, series bible, or game system — the way you select, arrange, name, and narratively contextualize them — likely IS protectable.

4. Thaler v. Perlmutter — Current Status

The D.C. Circuit affirmed the district court's ruling in March 2025, holding that AI cannot be an author under the Copyright Act. Thaler filed a cert petition with the Supreme Court in October 2025. The DOJ filed its opposition brief in January 2026, recommending the Court decline to hear the case. Most observers expect cert to be denied, which would make the "no AI authorship" rule effectively settled law in the federal circuits. This does not mean AI-assisted works are unprotectable — it means works generated solely by AI, with no meaningful human creative input, cannot receive copyright registration.

Bottom line for worldbuilders: the raw outputs of an AI prompt are likely not copyrightable on their own. But your creative work building on, integrating, and expressing those elements in your fiction almost certainly is. The question is always about the degree and nature of your human contribution.

This is exactly the thread I have been looking for. Here is my specific situation and I would appreciate any guidance.

I have been developing a fantasy series for about 20 years. The world has four moons, a 600-day orbital year, and unique tidal and seasonal patterns that affect the cultures and religions I have built. I want to use Claude to help me with the hard science — calculating how the four moons would actually interact gravitationally, what the tidal patterns would look like, how seasons would work with a 600-day year and the planet's axial tilt, and then building a working calendar system around those constraints.

The concept — four moons, 600-day year, the cultural significance of the tidal convergences — that is all mine, developed long before I ever touched an AI tool. But I want Claude to do the math and help me structure a calendar that is astronomically consistent. I would then name the months, assign cultural holidays, create the mythological framework around it, and weave it into the novels.

Is the resulting calendar system mine? If someone read my book and reproduced my calendar in their own work, would I have any recourse?

@SciFiNovelWriter — Your scenario is actually one of the stronger cases for copyright protection among the AI-assisted worldbuilding examples we see. Here is why.

What favors your position:

• The creative concept is entirely yours. The four-moon system, the 600-day year, the idea that tidal convergences have cultural significance — those are your original creative ideas, developed independently over two decades. Claude is not generating the creative vision; you are.
• Claude is functioning as a calculation tool, not a creative author. Asking Claude to compute gravitational interactions and tidal patterns based on your parameters is analogous to using a scientific calculator or a physics simulation. The tool performs the computation; you provide the creative inputs and make all the decisions about what to compute and how to use the results. No one would argue that using a TI-89 to calculate orbital mechanics makes Texas Instruments a co-author of your novel.
• You are making all the expressive choices. Month names, cultural holidays, mythological frameworks, how the calendar integrates into your narrative — those are classic elements of creative expression. That is where copyright lives.
• The compilation is clearly human-authored. Even under the most conservative reading of Zarya of the Dawn, your selection and arrangement of these elements into a coherent fictional world, integrated into novels, is protectable as a compilation.

Practical recommendations:

• Keep a detailed creative log showing which elements you conceived independently (the four-moon concept, cultural significance, character arcs tied to the calendar) versus what Claude calculated (orbital periods, tidal math, season lengths)
• Save your Claude conversation logs — they are evidence of your creative direction and the AI's computational role
• Use Claude Pro or the API rather than the free tier, so you have clearer contractual rights and can opt out of training data usage
• Document your iterative process — show that you revised, rejected, and reshaped Claude's outputs based on your creative judgment

As for recourse if someone copies your calendar: the novel containing the calendar is clearly copyrightable. The specific creative expression of the calendar — month names, cultural framework, narrative integration — is protectable. Someone who lifted your entire calendar system and its cultural context out of your novel would be infringing on your creative expression, not on the mathematical calculations underneath.

Where it gets weaker: if someone independently used similar astronomical parameters and arrived at a similar mathematical structure, you could not stop them. Facts and mathematical truths are not copyrightable. But the creative expression layered on top of those facts is yours.

Related question on a different worldbuilding element. I am building a constructed language for my novel. I designed the phonological system myself — specific consonant clusters, vowel harmony rules, an agglutinative morphology with case markers I invented. I then asked Claude to generate a base vocabulary of about 500 words and a set of grammatical rules that are consistent with the phonological constraints I specified.

Claude did a surprisingly good job. The words "feel" right for the sound system I designed, and the grammar hangs together. But Claude generated the actual words and the specific grammatical structures. I have since expanded the vocabulary myself, written poetry and dialogue in the language, and created an in-world dictionary for my series bible.

Who owns the language? If my book is published and someone uses my constructed language in their own fan fiction or derivative work, do I have any IP claim?

@ConlangCreator — This is one of the trickier areas in worldbuilding IP because constructed languages sit at an uncomfortable intersection of several legal principles.

The baseline problem: languages are not copyrightable.

No one can copyright English, or Mandarin, or Esperanto. A language is a system of communication — a method, a process — and under 17 U.S.C. 102(b), copyright does not extend to "any idea, procedure, process, system, method of operation." This principle was central to the Oracle v. Google API copyright dispute, where the Supreme Court analyzed how far copyright extends into functional systems.

But constructed language materials can be copyrightable.

There is an important distinction between "the language" and "the creative works documenting and expressing the language." The Tolkien Estate does not own "Elvish" as a language system, but they hold copyright in Tolkien's dictionaries, grammars, and — crucially — the creative texts written in those languages. Similarly, the creators of Klingon and Dothraki have navigated this distinction. The language system itself may not be protectable, but the specific creative expression — the dictionary entries, the poetry, the dialogue, the grammatical treatises — can be.

How the AI element affects your specific case:

• The phonological rules you designed are your original creative framework — but as a "system," they are likely not copyrightable on their own
• The 500 words Claude generated following your rules are AI outputs — copyright protection for those specific words is weak
• The vocabulary you expanded yourself, the poetry and dialogue you wrote in the language, the dictionary you compiled — those are human-authored creative works and are much stronger candidates for copyright
• The compiled dictionary as a whole (your selection, arrangement, and creative choices in documenting the language) may be protectable as a compilation, even if individual AI-generated entries are not

One additional consideration: if you asked Claude to "make up a language" with no further specification, your position would be substantially weaker than what you have described. The fact that you designed the phonological system and Claude generated words within your creative constraints is a meaningfully different scenario. You are closer to the "AI as tool executing human creative direction" end of the spectrum.

Practical advice: clearly separate your creative contributions (phonological design, cultural context, poetry, narrative use) from Claude's raw outputs (initial word lists, generated grammar tables) in your documentation. The stronger your paper trail showing human creative direction, the better.

Glad this thread exists. Maps are the one worldbuilding element where the AI question feels most immediate because the output is visual and directly published.

Here is what I did for my current project: I used Midjourney to generate a set of continent shapes and coastlines — basically the physical geography. I ran about 40 generations, selected 3 that had the right "feel" for my world, then composited them in Photoshop. From there, I hand-drew all the political borders, placed every city and labeled them, designed the trade routes based on my world's economic system, added mountain ranges and rivers where they made geographic sense, and created the legend and cartographic style.

The final map is about 70% my work by visual area, but the foundational landmass shapes came from Midjourney. Where is the ownership line here? If this map appears in the front of my published novel, what is and is not protected?

@MapMakerDave — Your scenario maps closely (no pun intended) onto the Zarya of the Dawn framework, which is helpful because it gives us a reasonably clear template.

What is likely NOT copyrightable:

• The AI-generated continent shapes and coastlines from Midjourney, considered in isolation. These are analogous to the individual AI-generated images in Zarya — you provided a text prompt, but Midjourney made the specific expressive choices about the shapes. Your selection from 40 generations shows curatorial judgment, but the underlying shapes themselves are AI-generated visual elements.

What IS likely copyrightable:

• Your hand-drawn political borders, city placements, and labels — these are your original creative expression
• The trade routes you designed based on your world's economic logic — creative choices driven by your fictional framework
• Mountain ranges and rivers you added — your geographic design decisions
• The cartographic style and legend — original artistic expression
• The compiled map as a whole — your selection and arrangement of all these elements (AI-generated and human-created) into a single cohesive work is protectable as a compilation

Practical implications: If someone reproduced your entire map, you would have a strong infringement claim based on the compilation copyright and the substantial human-authored elements. If someone traced only the coastline shapes and created an entirely different political/cultural map on top of them, your claim would be much weaker for those specific elements.

My strong recommendation: Keep your Photoshop layers separate and save them. Maintain the AI-generated base layer as a distinct file and your human additions as separate layers. This gives you clear evidence of which elements are yours. If you ever need to register the copyright or enforce it, being able to demonstrate exactly what you contributed is invaluable. The Copyright Office may require you to disclaim the AI-generated portions, similar to how the Zarya registration was amended, so having clean separation makes that process straightforward.

Chiming in from the publishing side. I am an acquisitions editor at a mid-size SFF imprint, and I want to share what we are seeing in the industry because it directly affects how writers should think about AI-assisted worldbuilding.

Current state of publisher AI policies (as of early 2026):

• Penguin Random House updated their submission guidelines in late 2025 to require authors to disclose any use of AI tools in the creation of their manuscript, including worldbuilding, research, and developmental stages. They do not prohibit AI use outright, but they want full transparency.
• Tor/Macmillan has a similar disclosure requirement. Their internal guidance distinguishes between "AI as research tool" (generally acceptable) and "AI as prose generator" (requires closer scrutiny and may affect contract terms).
• Several major agencies have added AI disclosure clauses to their author representation agreements. If you are querying agents, expect to be asked.

What this means for worldbuilders specifically:

Using AI to calculate orbital mechanics, check linguistic consistency, or model weather patterns is generally treated the same as using any other research tool. Nobody cares if you used Wolfram Alpha to check your planetary science — using Claude for the same purpose is viewed similarly by most editors I have spoken with.

The red flag is when AI-generated prose ends up in the final manuscript. If Claude wrote a passage describing your calendar system and that passage appears verbatim (or near-verbatim) in your submitted novel, that is where publishers get uncomfortable. It is not necessarily a deal-breaker, but it changes the conversation about the contract's warranty and indemnification clauses.

My advice to authors: use AI for the background work (calculations, consistency checks, brainstorming), but write every word of the final manuscript yourself. And when you submit, be transparent about your process. Publishers are far more forgiving of honest disclosure than of discovering undisclosed AI use after the fact.

Different use case here — I am designing a tabletop RPG and I used Claude extensively to help me develop the magic system mechanics. I described the flavor I wanted (elemental magic tied to seasonal cycles, with a risk/reward mechanic based on lunar phases), and Claude helped me design the actual rule framework: dice pools, modifier tables, spell progression curves, and balance math.

I then wrote all the lore, the spell descriptions, the flavor text, and the rulebook prose myself. But the underlying mechanical system — the numbers, the dice math, the balance structure — that was heavily AI-assisted.

Is this a different analysis from the novel scenarios discussed above? TTRPG publishing has its own IP norms (the OGL debacle, Creative Commons, etc.) and I am not sure how AI-generated mechanics fit into that landscape.

@GameDesignerLex — Good question. The analysis is actually somewhat different for game mechanics, and in a way that may work in your favor.

The key principle: game rules and mechanics are generally not copyrightable.

This goes back to Baker v. Selden (1879) and was reinforced in Lotus v. Borland (1st Cir. 1995). Game mechanics — the rules of play, the mathematical systems, the procedures — are "methods of operation" or "systems" under 17 U.S.C. 102(b) and are excluded from copyright protection. This is why anyone can make a chess game, a trick-taking card game, or a d20-based RPG system without infringing anyone's copyright.

What IS copyrightable in a game:

• The creative expression used to describe and explain the rules (your rulebook prose)
• The lore, narrative, and flavor text
• Artwork and graphic design
• The specific creative selection and arrangement of rules into a coherent game product
• Character names, place names, and other fictional elements

How this applies to your AI-assisted mechanics:

Because the mechanics themselves (dice pools, modifier tables, balance math) are likely not copyrightable regardless of who or what created them, the question of whether AI generated them matters less than it would for copyrightable creative expression. Nobody could copyright those mechanics even if a human designed them entirely by hand. Where AI authorship matters for copyright purposes — the expressive elements — you wrote yourself.

So paradoxically, your situation may be cleaner than the novel-writing scenarios: the parts AI helped create (mechanics) were never copyrightable, and the parts that are copyrightable (lore, spell descriptions, rulebook prose, flavor text) are human-authored.

One caveat: while mechanics are not copyrightable, they can potentially be patented (though game mechanic patents are rare and narrow). And trade dress protection may apply to the overall "look and feel" of your game product. But for straight copyright analysis, you are in a reasonable position.

Updated Summary: Practical Checklist for Fiction Writers Using AI Worldbuilding Tools

Based on the excellent analysis in this thread, here is a consolidated checklist for writers, game designers, and worldbuilders who are using AI tools in their creative process.

Documentation practices:

1. Maintain a creative decision log. Separate your original creative ideas (world concept, cultural frameworks, character motivations, narrative themes) from AI-generated outputs (calculations, raw word lists, generated text). Date your entries. This is your evidence of human authorship.
2. Save your AI conversation logs. Your prompts demonstrate creative direction. The back-and-forth shows iterative human judgment. These are valuable records if you ever need to demonstrate the nature of your contribution.
3. Keep iterative drafts. Show how you revised, rejected, and reshaped AI outputs. A first draft that is 50% AI-generated and a final draft that is 95% rewritten by you tells a compelling story about human authorship.

Usage practices:

1. Use AI as a research and calculation tool, not a prose generator. The strongest IP position is when AI handles computation (orbital mechanics, linguistic consistency, mathematical modeling) and you handle all creative expression.
2. Consider Claude Pro or the API for serious projects. Paid tiers offer clearer contractual rights and more control over whether your inputs are used for training. For detailed analysis of Claude's terms, see: Who Owns Claude's Outputs and How Can They Be Used?
3. Write every word of your final manuscript yourself. Use AI for background research and worldbuilding scaffolding, but ensure the published prose is human-authored.

Publishing and disclosure:

1. Disclose AI use to publishers per their current guidelines. Transparency is increasingly expected and protects you contractually.
2. Distinguish between AI-as-research-tool and AI-as-content-generator in your disclosures. Publishers treat these differently.

The overarching principle: The more your creative vision drives the process — the more you conceive, direct, select, arrange, revise, and express — the stronger your IP position. AI is a tool. The authorship is yours to the extent you actually exercise creative judgment throughout the process.

Key resources:

• Who Owns Claude's Outputs and How Can They Be Used?
• MEGATHREAD: AI-Generated Content, Copyright & Commercial Use
• Claude AI terms for business use
• AI-generated content copyright ownership

I will continue updating this thread as new developments arise. If you have a worldbuilding-specific AI ownership question, post it here.

This thread is incredibly valuable — thank you to the attorneys and everyone who contributed. I have been worrying about this for months and the checklist alone is worth bookmarking.

One more question that I think is on a lot of writers' minds: the Bartz v. Anthropic class action just settled for $1.5 billion. That case was about Anthropic training Claude on copyrighted books without authorization. Does that settlement change anything for those of us using Claude's outputs in our fiction? I have seen some anxiety in writer communities that the settlement somehow taints or undermines anything created with Claude.

@NovemberNovelist — The anxiety is understandable but largely misplaced. Let me explain why.

What Bartz v. Anthropic was about: The case concerned whether Anthropic's use of copyrighted books to train Claude constituted infringement. The $1.5 billion settlement resolved the claims of the plaintiff class (authors whose works were allegedly used in training data) without a ruling on the merits. It was essentially a licensing fee paid retroactively — Anthropic paid to settle the claim that it should have licensed the training data.

What it was NOT about: The settlement did not address the copyright status of Claude's outputs, the rights of end users, or the ownership of AI-assisted creative works. Those are entirely separate legal questions.

Does it affect end users? Not directly. The settlement resolved a dispute between Anthropic and the authors whose works were used for training. Your rights under Anthropic's terms of service — including your contractual right to use and commercialize Claude's outputs — are unchanged. The theoretical secondary liability risk I have discussed in other threads (the possibility that end users could be liable if AI tools were trained on infringing data) is actually reduced by the settlement, because the training data claims have been resolved. Anthropic effectively obtained a retroactive license.

What the settlement does signal is that the legal landscape around AI and creative works is maturing. Companies are negotiating licensing frameworks. Courts and parties are finding resolution paths. That is broadly positive for everyone in the ecosystem, including end users.

So no — the Bartz settlement does not taint your Claude-assisted worldbuilding. Keep following the best practices outlined in this thread, document your creative process, and focus on making your fiction as compelling as your world deserves.