Fanfiction Writers Fight Back as AI Scrapes Their Stories for Profit

Fanfiction Writers Fight AI Scraping, One Story at a Time

Fanfiction writers create stories inspired by beloved movies, books, and games, sharing their work freely within supportive communities. A key unspoken rule in these spaces is to never charge for fanfiction or steal others’ work. Given this, many fanfic writers were among the first to raise concerns when their work was used without permission to train AI language models.

In early April, a user named nyuuzyou scraped 12.6 million fanfics from Archive of Our Own (AO3), a major fanfiction repository, and uploaded the dataset to Hugging Face, a platform hosting open-source AI models. The fanfic community quickly discovered this and reacted with outrage, creating tools like the ao3scrapesearch to help authors check if their work had been scraped.

Many writers expressed their anger and frustration in the comments on Hugging Face and elsewhere online. Some defended the scrape, arguing that large tech companies’ bots had already gathered this data multiple times. But others saw this as exploitation of their creativity and labor.

The Heart of Fanfiction Communities

Nikki, a Star Wars fanfic writer known online as infinitegalaxies, found over 70 of her stories scraped, including a collective essay she co-wrote about AI’s threat to fandom. Nikki primarily writes Reylo fanfiction, centered on the romantic pairing of Rey and Kylo Ren. The Reylo fandom is vibrant and diverse, with tens of thousands of stories exploring everything from canonical battles to coffee-shop romances and medieval fantasy.

Fanfiction communities like Reylo offer more than storytelling—they provide a sense of belonging and creative freedom. Writers like Nikki and Em (pen name okapijones) cherish the community and the absence of traditional publishing restrictions. Em highlights how fanfic is a "pure creative playground" that fosters innovation without editors or rules.

Why AI Scraping Feels Like Theft

Fanfic writers invest time, effort, and emotion into their work, often with no financial gain. Their stories are gifts to a community. Seeing AI scrape and repurpose their work without consent feels like a violation.

Nikki recalls the shock of discovering AI tools like Sudowrite’s Story Engine, which can generate novels based on minimal prompts. The realization that AI could copy fanfiction content without permission made her emotional. In 2023, AI models even began reproducing specific sexual terms from niche fanfic genres, confirming unauthorized training on fanfiction data.

Many in fandom view AI-generated fanfiction or fanart as theft, lacking ethical justification when built on stolen labor. Unlike fanfiction, which is non-commercial and transformative, AI models merely remix existing content without true creativity or consent.

Ongoing Battles Against AI Exploitation

Fanfic writers have seen repeated attempts by AI entrepreneurs to monetize their work. For instance, Cliff Weitzman, CEO of Speechify, scraped thousands of AO3 fics for a linked website but removed them after fan backlash. Another app, Lore.fm, marketed itself as "Audible for AO3" before shutting down due to community opposition.

Writers feel like they are playing whack-a-mole, constantly defending their work against new attempts to exploit it. Despite using copyrighted characters as a creative sandbox, fanfic creators hold rights to their original expressions and stories, as AO3 explains. This is fundamentally different from how AI models use data.

Community Responses and Legal Actions

Some fans have tried generating new chapters with AI, frustrating authors. Others restrict access to their stories, requiring AO3 accounts or removing content entirely. After nyuuzyou’s scrape, fans and the Organization for Transformative Works (OTW), which runs AO3, filed takedown notices under the DMCA. Hugging Face disabled the dataset on April 9.

OTW has implemented protections like CloudFlare to block AI scraping, though these measures aren’t foolproof. Nyuuzyou, an 18-year-old student in Russia, defended his scrape as “legitimate research” unrelated to commercial AI development, and reuploaded the data to platforms less responsive to takedown requests.

The Bigger Picture on AI and Fanfiction

Hugging Face aims to democratize AI by hosting open-source models and datasets, with backing from major companies like IBM. But experts remain skeptical of scrapers claiming their data won’t be used for AI training. After all, large unstructured datasets are primarily valuable for building language models.

Alex Hanna, a research director at the Distributed AI Research Institute, urges sites like AO3 to continue aggressively protecting creators’ work. For fanfic writers like Nikki and Em, the fight against AI scraping is personal. They create freely, out of love for the characters and community, only to see their efforts taken without permission.

“I don’t go looking for a fight,” Nikki says. “But when people come to us with a fight, I will fight.”

What Writers Can Take Away

• Be aware that AI scraping of creative work is a real issue affecting many online communities.
• Understand that fanfiction writers invest personal time and creativity without financial gain, making unauthorized use of their work especially harmful.
• Support platforms and communities that protect creators and push back against unauthorized data scraping.
• Consider legal avenues and community action when your creative work is used without permission.

For writers interested in learning more about AI’s impact on creativity and how to responsibly engage with AI tools, resources like Complete AI Training offer practical courses and guidance.