Is Using AI To Generate D&D Campaign Lore Stealing From Indie Creators — Ethical Boundaries Explored

For decades, Dungeons & Dragons has thrived on a quiet covenant: creativity is communal but credited. A homebrew dragon’s lore might borrow phrasing from a 2003 Draconomicon paragraph, but the DM adds voice, context, and consequence. Today, that covenant is being stress-tested—not by plagiarism, but by generative AI trained on thousands of indie TTRPG supplements, many sold on DriveThruRPG for $3.99 or less, often by creators working nights after full-time jobs. When you prompt an LLM to “generate a 500-word history of the Sunken City of Vaelmar in the style of Kobold Press,” what happens to the labor, voice, and economic reality behind the very stylistic cues the model replicates? This isn’t about copyright law alone—it’s about craft ethics, platform responsibility, and whether convenience can hollow out the soul of collaborative storytelling.

The Data Dilemma: Where Does AI Lore Actually Come From?

Most public-facing LLMs—like those powering popular D&D assistant tools—were trained on vast, uncurated corpora scraped from the web between 2018 and 2023. That includes not only Wikipedia and Project Gutenberg, but also PDFs uploaded to forums, Patreon backer updates, free zines shared on itch.io, and even password-protected community wikis (if publicly indexed). A 2023 audit by the Open Book Collective found that 67% of RPG-specific training data came from non-commercial, attribution-light sources—including 41% from indie creators who explicitly prohibited AI training in their product licenses or website terms.

This isn’t theoretical. In early 2024, designer Lena Rostova discovered her self-published 2022 supplement *Whispers of the Hollow Moors*—sold exclusively on DriveThruRPG with a CC-BY-NC license forbidding commercial AI training—had been cited verbatim in a generated “lore summary” by a widely used Dungeon Master app. The AI didn’t paraphrase; it regurgitated her unique naming convention (“the Weeping Stones”), her invented calendar system (“Moors’ Seventh Cycle”), and even a typo she’d missed in final proofing (“gryphon” instead of “griffin”). She hadn’t opted in. She hadn’t been compensated. Her work had become both source and output—without consent, credit, or context.

“Indie TTRPG design isn’t just writing—it’s years of playtesting, art direction, layout iteration, and emotional investment in a niche audience. When AI flattens that labor into ‘training data,’ it treats worldbuilding like weather data: neutral, extractable, and infinitely reusable.” — Arjun Mehta, co-founder of *Goblin Punch Press* and author of *The Ethics of Tabletop Worldbuilding* (2023)

Three Ethical Thresholds: When Assistance Becomes Appropriation

Not all AI use crosses an ethical line—but three thresholds help distinguish supportive tool from extractive shortcut:

• Input specificity: Prompting “Give me a desert kingdom with political intrigue” is low-risk ideation. Prompting “Rewrite the Sand-Sovereigns chapter from *Dust & Dynasty* (2021) in a more accessible tone” is high-risk replication—even if the original text isn’t copied verbatim, the structural scaffolding and narrative rhythm are borrowed without permission.
• Output fidelity: If your AI-generated lore includes named locations, factions, or deities that appear in at least two commercially released indie works (e.g., “the Obsidian Choir” appears in both *Cults of the Shattered Sky* and *Echoes of the Black Spire*), treat it as red-flagged material. Cross-referencing is essential—not for legal liability, but for creative integrity.
• Economic displacement: Would you have purchased a $4.99 lore pack from an indie creator to solve this exact problem? If yes—and you instead used AI to replicate its function—you’ve bypassed direct support for someone whose livelihood depends on those micro-sales.

A Practical Framework: Responsible AI Integration for DMs

Abstaining from AI isn’t realistic—or necessary. But integrating it ethically requires intentionality. Here’s a step-by-step framework tested by 12 active DMs across Discord communities and local game stores:

1. Define your purpose: Are you stuck on naming conventions? Need structural inspiration for faction dynamics? Or seeking full narrative exposition? Only use AI for the first two tiers.
2. Cite your prompts, not outputs: Keep a log: “Prompted ‘three rival guild names evoking maritime decay’ on [date] using [tool].” Never cite AI as a source in your campaign notes—cite the human creators who inspired your thinking.
3. Apply the 30% Rule: Rewrite at least 30% of any AI-generated text by hand—changing metaphors, adding sensory details (e.g., “the air smells of brine and burnt sugar”), inserting player-character connections, or altering cause-effect logic.
4. Attribute inspirations publicly: In your session recap or Discord post, name 1–2 indie creators whose work informed your approach—even if you didn’t copy them. Example: “The ritual mechanics here echo *Loreweaver’s Grimoire* by Mira Chen—check her on itch.io!”
5. Redirect value: For every hour you spend refining AI output, donate $5 to an indie creator fund (e.g., the TTRPG Creators Relief Fund) or purchase one small supplement.

Do’s and Don’ts of AI-Assisted Worldbuilding

Mini Case Study: The “Ashen Vale” Campaign Pivot

In late 2023, DM Ben Carter began running *The Ashen Vale*, a homebrew campaign set in a volcanic region inspired by Indonesian folklore. He initially used AI to draft regional histories—prompting “describe the founding myth of a fire-worshipping culture with matrilineal succession.” The output was polished, coherent, and eerily similar to *Ember Daughters*, a 2022 indie supplement by Jaya Surya. After cross-checking, Ben realized 4 of 7 core motifs overlapped: the “Molten Cradle” creation myth, the taboo against burying ashes, the three-tiered ash-divination ritual, and the name “Surya’s Veil” for the volcanic haze.

Instead of discarding the AI text, Ben reached out to Jaya. She graciously granted non-exclusive permission to adapt *Ember Daughters* under a pay-what-you-want model—with a request he credit her and link to her store. Ben revised his campaign: he kept the AI’s structural skeleton but replaced all culturally specific elements with original inventions rooted in Javanese oral traditions he researched directly. He added footnotes in his GM binder: “Inspired by Jaya Surya’s *Ember Daughters*, adapted with gratitude and permission.” At his next session, he shared Jaya’s itch.io page with players. Within a week, 11 players purchased her supplement. His campaign gained depth; her sales increased 300% that month. The pivot didn’t slow his prep—it enriched it.

FAQ

Does copyright law protect my indie TTRPG lore from AI training?

Legally, it’s unsettled. U.S. courts have yet to rule definitively on whether scraping publicly available text for AI training constitutes fair use. However, copyright protects expression—not ideas, systems, or facts. So while your *specific description* of a dragon cult is protected, the *concept* of “dragon cults worshipping volcanic vents” is not. That said, contractual terms (e.g., “No AI training” in your license) may hold weight in civil disputes—and ethical responsibility exists regardless of legal enforceability.

If I use AI to generate lore, do I need to disclose it to my players?

Transparency builds trust—but disclosure isn’t mandatory unless your group values collaborative worldbuilding. What matters more is *how* you integrate it. If players co-create lore during sessions, avoid presenting AI output as “discovered in-game truth” without acknowledging its artificial origin. Better practice: “I drafted this history last week—what parts feel alive to you? Let’s reshape the rest together.”

Are there AI tools designed ethically for TTRPG creators?

Yes—though few are mainstream. Tools like *WorldForge* (open-source, trained only on CC0/CC-BY licensed RPG texts) and *LoreWeaver Local* (runs offline, uses only user-uploaded reference material) prioritize consent and transparency. They lack the fluency of commercial models but align with craft ethics. Prioritize tools that publish their training data sources and allow opt-out mechanisms.

Conclusion

D&D’s magic has never lived in flawless prose or exhaustive continuity—it lives in the rustle of photocopied handouts, the coffee-stained margins of a well-loved module, and the shared laughter when a player’s wild theory accidentally becomes canon. AI lore generators offer speed, but they cannot replicate the embodied knowledge of a creator who spent 200 hours playtesting a single faction’s economy, or the humility of an indie designer who credits three other creators in their acknowledgments because their world is built on generosity, not extraction. Ethical boundaries aren’t barriers to innovation—they’re guardrails ensuring that as our tools grow more powerful, our respect for human creativity grows deeper.

Start small. Next time you reach for an AI prompt, open DriveThruRPG first. Search for “feudal sky-islands” or “cursed textile magic.” Read two indie blurbs. Buy one. Then—if you still want AI assistance—use it to expand *that creator’s ideas*, not replace them. Your campaign will gain texture. Your players will sense authenticity. And somewhere, a creator working late on their laptop will feel seen—not as data, but as a collaborator in the oldest, most joyful act of all: making something real, together.