In early April, a user on Hugging Face, under the nickname nyuuzyou, uploaded a collection of 12.6 million fanfics, which were gathered from the Archive of Our Own (AO3) platform without the knowledge or permission of the authors. This sparked a wave of outrage in the fan community, which quickly created a special search engine to detect stolen works.
Fanfiction has long operated on the principle of a gift economy — authors do not profit and share their works solely for the enjoyment of readers. Many see the use of their stories for AI training as a violation of ethical norms and direct theft. The outrage was further fueled by previous attempts by companies to profit from fanfics — for example, when Speechify uploaded thousands of fanfics to its site without permission, or when the Lore.fm app was advertised as “Audible for AO3” and was forced to shut down after protests.
After the discovery of the large fanfic leak, fans organized an online campaign to file DMCA complaints, joined by the non-profit Organization for Transformative Works, which manages AO3. On April 9, Hugging Face removed the dataset, and AO3 strengthened its protection against automated data collection, although it acknowledged that a perfect solution is not yet available. However, nyuuzyou remained steadfast and re-uploaded the dataset on overseas servers, despite the community’s outrage.
Nyuuzyou himself stated that he acted with research purposes in mind and is not interested in fanfiction, claiming his goal is to support the development of tools for content moderation and archive preservation. However, many authors and experts doubt the sincerity of such explanations and believe that such collections are inevitably used to train language models without the creators’ consent.
Fanfic authors are not giving up and continue to fight against AI intrusion into their community. They are convinced that their creativity should not become a resource for automated systems and are using all available means to protect their works from unauthorized use.
