Archive Of Our Own is a website with a specific mission: serve as an online archive for written content that other websites won’t touch anymore, because advertisers do not want their ads next to a lengthy politically-dramatic retelling of the My Little Pony series, and most websites today rely on ads for server funding. AO3 is different – it runs on a combination of volunteer work and donations. While largely used for fanfiction, it has a small community of original fiction writers and visual artists as well. Unfortunately, as you might expect, the archive doing its job as an archive upsets some people badly enough that they’ll go out of their way to try and make problems for everyone using it.
Negative Nancy-Bot
The true reason anyone leaves a negative comment on a fanfiction, a largely harmless body of fictional text that was provided for free, tagged appropriately, and posted to a site that forces an age check on fics that might be too scary, are unknowable. Bad day? Secret expectations for a given work that this fan, who’s age might jump anywhere between “old enough to have used a typewriter” and “too young to have ever used a physical keyboard”, did not meet? Who knows. The community of AO3, which contains members who have watched every single other fanfiction site implode, is largely very welcoming, and you only very rarely see negative comments come from a registered account (which you have to actively ask for, on a waitlist, unlike most other websites). Guest accounts are another story.
With the rise of AI-generated text came, bizarrely, AI-generated hate comments. Where the previous hate comments were explainable as a “bad day” or someone who thinks they’re the funniest, most sarcastic person on the planet, the new ones generated by ChatGPT or similar AI are just… kind of… bizarrely mean. Prior generations were as simple as “I ran this text through an AI detector and it says you’re a dirty AI user lying to pretend you’re an author, you should delete this and also everyone hates you nyeh”. This is noteworthy, because the person running this script to leave this comment in multiple places is also using AI, which makes this seem like trolling anti-AI communities. That’s annoying, but understandable. Someone could do that, even if there’s no reason to. Similar comments attempted to insult the author by telling them their work sucked and they’d die alone, but given the state of discourse around AO3, those were virtually background noise, and less of those got posted to social media like BlueSky or Tumblr telling other authors to watch out. It was just sort of understood that certain types of fiction would attract those comments.
But the next generation, after multiple fan fiction writers got together and realized they had similar or even identical hate comments on their work, was scary: it specifically cited characters contained in the work, attempted to cite the tags on the work, maybe even used literary terms like ‘plot twist’ or ‘purple prose’. They were much harder to weed out; author groups were struggling to find a pattern outside of the generic “this sounds like AI”, and the unfortunate conclusion had to be that someone was putting in a bunch of work just to leave mean comments under fan fiction. The alternative would have been insane. A large number of authors started getting these all at the same time; it would have had to have been a coordinated campaign by some other group, and maintained over months, slowly improving coincidentally with every new release of ChatGPT, or a machine-led effort. There’s clearly script-work going on, so this isn’t the same as someone reaching a part of the story that disturbed them, leaving a knee-jerk comment telling the author to kill themselves on a guest account, and then leaving. This is premeditated.
Why Even Bother?
Theories as to why someone with the expertise to set up a bot to A) go to ChatGPT (or similar) and ask it to “read” the fic, B) have the AI write an extremely mean comment diminishing the author as a person, or threatening to doxx them, or accusing them of using AI, etc. C) and then run through AO3’s Guest Comment setup to leave the response ChatGPT (or similar) then gives would do all that were understandably confused. Is this someone trying to get the author to delete the work, so that they (the bot’s creator) could then… post the same fic somewhere else, now that the original’s gone, so they can pretend they own it and wrote it? Well, the problem is that AO3 allows for works to be “orphaned”, so if an author just didn’t want an old work associated with them anymore, they can simply sever that tie without deleting it. Are they trying to copy the author’s writing style into their AI generator? Sure, they might be trying to do that. But. Why? They already scanned the work to get the mean comment, so it’s already in there, right? Why do they also need the author to delete the fic if they plan to generate new content? They’d have to go beyond just getting an individual fic deleted, they’d have to hope the author deletes their entire account and never comes back, ever. AO3’s community is tightly-knit enough that just firing these off at random won’t magically give some nefarious hacker a digital skin-mask of a really popular person to wear. Someone will notice. You can’t just sneak into a community with the same writing style and an almost identical name to an author who may have been complaining about bots somewhere the troll can’t see for months. And there’s also the question of why you’d even want to impersonate someone in the first place, because you can’t do monetary transactions on AO3 because it’s one of the few things AO3 explicitly forbids users from doing. Is it purely to phish people? Are they hoping to impersonate a familiar author to phish people? On AO3? In DMs? On the incredibly low chance they’re successful? Genuinely, what purpose does this even serve?
The easiest answer is that it’s simple hatred for the craft, but that seems too simple for the amount of work this takes, until you consider the type of person who just hates fanfiction. You’re looking at someone who doesn’t want to search for anything in particular because the things they hate will be there, is probably disgusted by slash fiction, and likely doesn’t understand why anyone’s writing “fix-it” stories. Having a machine just troll through newer submissions at random may be less mentally distressing for the type of person doing all the work this would take. They don’t have to do it at all, of course, but feel compelled to.
And that hatred can run deep. AO3 does contain some works that are very dark, and they allow those on purpose because it’s the point of the archive: this is creative re-use of existing properties, done with no monetary gain. It’s not an exaggeration to say AO3 is defending the first amendment by doing this, a privilege that not many websites have. A lot of other websites might like to do the things AO3 does, but they cannot, because of where their funding comes from. AO3 itself is not beholden to advertisers, so this concept of “advertisers” that love stuff to be PG-E don’t get a say in what’s on the website, something almost every other popular fiction website must fight with constantly! People can run into stuff they’d never see anywhere else, but crucially, if they’re using the site right, they won’t run into it by accident, because everything is tagged. The tougher content even forces a “you sure you want to read this? It’s mature audiences only” page you have to click through before you can continue to the work in question. It does that even for works with nothing more than an “author chose not to tag this on purpose” tag.
It has to be pure hate for hatred’s sake. Maybe hatred of the website for allowing the darker content to stay, or maybe hatred for the authors who are continuing to use the site alongside authors writing thought experiments using characters from other works of fiction.
