Why fandom says no: censorship, AI, and the gift economy

Fandom's two fiercest values, resistance to censorship and rejection of generative AI, come from the same experience: repeatedly having its work deleted, policed, or taken without consent by platforms it didn't control. This page tells that story fairly, including where fans themselves disagree.

The purges: a community keeps losing its home

In 2002, FanFiction.net banned NC-17 content and removed enormous amounts of existing work; it purged again in 2012, deleting thousands of stories with little warning. In May 2007, LiveJournal, then fandom's social hub, mass-suspended hundreds of journals and communities in a panic over adult content, an event fans call Strikethrough (suspended journals appeared with their names struck out). A second wave, "Boldthrough," followed months later. In December 2018, Tumblr banned adult content and its fandom communities scattered overnight.

The pattern fans took from this: a platform that hosts your community for free can also delete it in an afternoon, usually to please advertisers or payment processors, and the sweeps always catch far more than their stated targets, including queer content, art, and discussion that broke no rules.

The response: an archive of their own

Weeks after Strikethrough, fans organized. In 2007 they founded the Organization for Transformative Works (OTW), a nonprofit with a pointed mission, and in 2008 opened Archive of Our Own, named after Virginia Woolf's A Room of One's Own. Its design principles are the purges inverted: fan-owned servers, donation funding (no advertisers to appease), maximum inclusiveness of legal content behind a clear rating and warning system, and a legal advocacy team that defends fanworks as transformative works. AO3's warning-and-tag system is the community's alternative to censorship: instead of a platform deciding what adults may read, writers label honestly and readers filter for themselves.

"Is this even legal?" Transformative works

Noncommercial fanfiction occupies a well-defended position in US law's fair use doctrine, which favors uses that are transformative, adding new meaning and expression rather than substituting for the original. Fanfiction comments on, extends, and reimagines its source; it isn't sold; and it doesn't replace the market for the original (nobody skips a novel because fic of it exists). No US court has ruled against noncommercial fanfiction as a category, and the OTW's legal team exists to keep it that way. The community's side of the bargain is the noncommercial norm: the moment fic is sold, the legal ground and the social contract both shift. That is why "don't sell fic" is one of fandom's few unbreakable rules.

Why the community rejects generative AI

When large language models arrived, fans discovered that fanfiction, millions of works posted as gifts, had been scraped into training datasets without anyone's knowledge or consent. Popular fics could be coaxed out of chatbots in recognizable fragments. For a community whose entire ethic is built on consent, credit, and labor freely given, this landed as a profound violation. The objections, specifically:

• Consent. Fans share work under an implicit social contract: read it, love it, don't take it. Scraping broke that contract at industrial scale, and "it was publicly viewable" persuades no one who lived through the purges.
• The gift economy. Fic is written for free and paid for in comments. AI-generated "fic" flooding archives takes from that commons without ever giving back. There's no author to talk to, no 3am reply to your comment.
• Labor and craft. A 300,000-word slow burn represents years of unpaid, loving work. Watching it become training data for commercial products feels, to many writers, like theft with extra steps.
• Trust. Fandom runs on pseudonymous trust between writers and readers. AI-generated works posted under human pretenses corrode it.

The practical effects are everywhere: AO3 added a preference for creators to block third-party AI scraping and does not permit its data to be licensed for AI training; fandom events routinely ban AI-generated entries; writers migrated drafts to tools like Ellipsus and Fileverse that pledge not to train on them; and "no AI" notes now appear in fic author's notes the way disclaimers once did.

Honesty requires saying: this isn't perfectly unanimous. You'll find fans who use AI privately for brainstorming and see no harm, and ongoing arguments about where lines sit. But the community consensus, visible in every archive policy, event rule, and comment section, is firmly against generative AI in fanworks, and a newcomer should understand that posting AI-generated fic will, in most spaces, be received somewhere between coldly and radioactively.

What this means for you, practically

• Don't feed fics to AI tools, not for summaries, not for "what should I read next," not to continue an unfinished WIP. To the author, all of it is taking.
• Don't repost or archive others' work anywhere, even with credit, without permission.
• Respect tags and ratings. They're the community's alternative to censorship, and they only work if everyone honors them.
• Curate your own experience. Filter out what you don't want to see instead of demanding it be removed. This norm is the entire settlement of fandom's censorship wars.