Fanfiction Writers Fight AI Scraping for Their Work
▼ Summary
– Fanfiction writers oppose AI using their work without permission, as seen when 12.6 million fics were scraped from AO3 and uploaded to Hugging Face for AI training.
– The Reylo fandom, a close-knit community, values creativity and free exchange, with many writers feeling their labor is exploited when AI uses their work.
– Fanfic authors like Nikki and Em argue AI-generated content violates fandom norms, calling it theft and a threat to their gift-based, community-driven culture.
– Despite pushback, scrapers like nyuuzou reupload datasets to evade DMCA takedowns, claiming “research purposes,” while critics doubt their motives.
– Experts like Alex Hanna suggest fanfic writers must continue fighting AI encroachment, as even small scrapes contribute to larger unethical AI training practices.
Fanfiction communities are rallying against unauthorized AI scraping of their creative works, with writers expressing outrage over datasets built from their stories without permission. These passionate creators, who share their work freely within tight-knit fandoms, view AI’s intrusion as both a violation of trust and a threat to their collaborative culture.
The conflict escalated when a dataset containing 12.6 million fanfics from Archive of Our Own (AO3) appeared on Hugging Face, a platform for open-source AI models. The upload, attributed to a user named nyuuzyou, sparked immediate backlash. Writers discovered their stories had been harvested through a search tool created by concerned fans, revealing the scale of the scraping.
For many authors, fanfiction represents more than just hobby writing, it’s a labor of love shared within supportive communities. Nikki, a Star Wars fanfic writer, found over 70 of her works in the dataset, including a collaborative essay warning about AI’s impact on fandom. The irony wasn’t lost on her. “This is something that takes time and effort and your heart and your soul,” she explains. Seeing it repurposed for AI training felt like a betrayal of the gift economy central to fan culture.
Platforms like AO3 operate on principles of transformative works, where writers adapt existing universes while retaining rights to their original contributions. Unlike AI-generated content, which recombines existing material, fanfiction involves human creativity, reimagining characters in new scenarios, from coffee shop romances to mafia dramas. The Reylo fandom alone boasts 30,000 stories exploring every conceivable variation of Rey and Kylo Ren’s relationship.
When AI tools like Sudowrite began generating content using terminology unique to fanfic tropes, writers recognized their work had been used to train these systems. “It’s theft at its core,” Nikki states. The issue gained further attention when apps like Speechify and Lore.fm attempted to monetize scraped fanfiction, retreating only after vocal fan opposition.
Hugging Face eventually removed the dataset following DMCA complaints, but nyuuzyou reuploaded it to servers in countries with lax copyright enforcement. Claiming academic intentions, the uploader argued the data could aid research. an explanation that rang hollow for critics. “Why collect unstructured text if not for AI training?” asks researcher Alex Hanna, who sees such scrapes as part of a broader pattern of unconsented data harvesting.
Fandom’s response highlights a fundamental clash between community values and AI development practices. Where fanfiction thrives on human connection and creative exchange, AI models treat stories as raw material. As Em bluntly puts it: “We do this for love, not profit. Seeing our work fuel systems with commercial ambitions feels like exploitation.”
While legal battles continue, many writers now restrict story access or delete works entirely, a defensive move that alters fandom’s traditionally open nature. For veterans like Nikki, the fight goes beyond individual datasets: “When people come for our community, we’ll push back. This isn’t just about copyright, it’s about protecting spaces where creativity flourishes.”
The standoff reflects larger debates about AI ethics and artistic labor, with fanfiction writers emerging as unexpected frontline defenders. Their struggle underscores how even non-commercial creative ecosystems face disruption when tech companies seek training data without regard for origin or consent. As Hanna observes, “Challenging each infringement might seem like whack-a-mole, but it’s currently the most effective resistance.”
(Source: The Verge)
