Fearing AI, fan fiction writers lock their accounts | TechCrunch

Kinktober. Whumptober. Kisstober. Flufftober. Goretober. October is a bacchanal of fan fiction, from romantic one-shots about unconventional character pairings to delicious smut that’ll make you reconsider your own sense of morality — all inspired by the month’s countless themed writing challenges. It’s an especially busy time for the fan fiction site Archive of Our Own (AO3). 

But this year’s monthlong prompt festival may seem quieter to the casual AO3 reader, with popular writers’ work seemingly wiped from the site altogether. In most cases, the stories still exist, but they aren’t publicly viewable anymore. 

In an effort to prevent their writing from being scraped and used to train AI models, many AO3 writers are locking their work, restricting it to readers who have registered AO3 accounts. Though it may curb bot commenters, it also limits traffic from guest users, which can be a blow for newer and less popular writers. Whether it’s effective is questionable, but in the AI paranoia, AO3 writers are taking any measures they can to protect their work. 

At the time of reporting, over 966,000 of the roughly 11.7 million works on AO3 were accessible only for registered users. It’s just a fraction of AO3’s vast library of content, but it’s worth noting that many authors are only locking new work, since existing fics were likely already scraped. 

Some readers took to Tumblr and Twitter to ask their favorite fan fiction authors if they had taken down their writing. One asked AO3 writer takearisk to unrestrict their work so they could read it on Tumblr, where many use RSS feeds to keep up with new chapters. 

“thanks so much for reading!! but no,” takearisk responded on Tumblr last week. “i had some ai bot comments a few months back that really freaked me out and i also found one of my older marvel fics posted to another site without my permission. it’s something that i never wanted to have to do, but i put way too much effort into my fics to be okay with them being stolen. so i am going to keep my account locked for the foreseeable future.”

The push to lock down AO3 began as early as last December, when ChatGPT and other generative AI tools began gaining popularity. 

It started when a Reddit user and AO3 writer found Omegaverse references in content generated by the controversial AI writing app Sudowrite. The Omegaverse is a speculative erotic fiction genre popular in fandom circles that revolves around wolflike mating dynamics between “alphas” (who impregnate others) and “omegas” (who are impregnated). The dynamic transcends assigned sex and gender; Omegaverse content often portrays male pregnancy (known as mpreg) and same-sex relationships. Breeding is performed through an act known as “knotting.” 

At the time, Sudowrite used GPT-3 to generate fiction content. Like many AI models, the program was trained using data scraped from the swaths of information available online. As the writer pointed out in a Reddit post, AO3 is one of the “largest and most accessible text archives” on the internet. Sudowrite, the writer posted, generated passages that not only mentioned Omegaverse terminology, but also demonstrated an understanding of the trope’s dynamics. 

Fan fiction writers later trolled AI generators by participating in a weeklong Omegaverse-themed writing marathon called Knot in my Name. Many writers with locked accounts still keep their Omegaverse content public in hopes of skewing future datasets with breeding references.

AO3 addressed the community’s AI-related concerns in a public announcement in May and suggested that writers restrict their work to registered users only in order to avoid data scraping. Doing so won’t block every potential scraper, the announcement said, but it “should provide some protection against large-scale scraping.” In addition to placing measures to hinder large-scale scraping, such as rate limiting, the site said it also implemented a code to opt out of Common Crawl, the web archive used to train generators like OpenAI’s ChatGPT. 

“Putting systems in place that attempt to block all scraping would be difficult or impossible without also blocking legitimate uses of the site,” the AO3 announcement said. “With that said, it is an unfortunate reality that anything that is publicly available online can be used for reasons other than its initial intended purposes.” 

Writers have been locking their accounts en masse since AO3’s announcement, despite reader requests to keep their work public. Registering for an AO3 account is cumbersome, since users have to wait for an invitation code. The site sends out 7,000 invitations a day, and at the time of reporting, there were over 40,000 people on the waiting list. Users may wait over a week to receive a code after they request it. Like Bluesky, the site may distribute invites to registered users.

“I know this seems like an extra step and maybe you don’t think you need it,” a Tumblr user said in a post imploring followers to register for AO3. “If you enjoy fics and you want to keep them coming, this is how you support your favorite writers! If our stats and comments plummet, I guarantee writing is going to start going down as well.” 

Many writers are also restricting their accounts to avoid the influx of bot comments, which became more common in the past year. Many are the typical scam comments, promoting porn sites, sketchy AI detection tools or the predatory fiction app Webnovel. It’s the comments that aren’t obvious scams that are raising AO3 users’ suspicions. 

The comments are generic, pleasant and don’t mention specific details about the story. They also don’t have a profile attached, which means they aren’t from registered users.

Some users speculate that data scraping tools automatically leave comments to “make their browsing traffic look legitimate,” according to one Tumblr user. AO3 users also speculate that the comments are being used to test AO3’s spam detection filters, or that they’re an attempt to encourage writers to keep their fics public and scrapable. AO3 authors have been accused of posting AI-generated fiction, or have expressed concern that the positive comments they leave on other writers’ fics will be misconstrued as AI spam. 

Whatever it is that’s driving the bot comments — nefarious or not — it’s fueling the AI panic among fan fiction writers. Some writers are keeping their accounts public, but have started adding disclaimers forbidding the use of their work in AI training. AO3 author notes are starting to include bold text notices like “I do not give permission for my fics to be copied and reposted elsewhere or fed to AI,” or “I do not give any permission for AI technology to copy my writing, or train themselves using my content.” Like the ancient Facebook privacy hoax that keeps coming back, public declarations will not ensure digital privacy. Companies like OpenAI are already notorious for using people’s personal data without their consent, and they’re unlikely to be stopped by a meager disclaimer. 

Locking AO3 accounts could protect new work from being included in training datasets, but it doesn’t stop the actual AI problem in fan fiction: other humans who use AI tools to generate endings for unfinished stories. Users have also posted about using AI tools if they didn’t like a fan fiction’s ending, or if an author took too long to post an update. They’ve also bragged about feeding fics to AI chat companions like Character.ai to feel like they’re interacting with their favorite characters.

“Not to sound like a boomer, but ai fucked how kids navigate the world, including fandom spaces,” Reddit user zoey1bm commented on a post warning other authors.

Reddit, TikTok and Tumblr are teeming with discourse over the ethics of feeding existing fics to AI tools. It may be legally fine, since copyright laws pertaining to AI and fan fiction are either nonexistent or do not favor writers, but the practice is largely considered a dick move because it involves adding someone else’s work to a database without their knowledge or consent.

By restricting their work to registered users, AO3 writers also sacrifice traffic and encouragement from anonymous guest users. Online, AO3 writers have posted about seeing a decline in views and comments since they locked their accounts. 

For others, the AI fears were short-lived. In recent weeks, Tumblr users have deliberated whether or not it’s “safe” to unlock their accounts, or questioned whether going private was effective in the first place. Others are resigned to whatever scraping is bound to come for them. 

“I, too, have unlocked my fics,” one writer said on Tumblr. “I figure at this point, it is what it is, even if I don’t like it.”