AI and Data Scraping on the Archive

Post Header

Published:

2023-05-13 21:58:58 UTC

Translations:

Tags:

With the proliferation of AI tools in recent months, many fans have voiced concerns regarding data scraping and AI-generated works, and how these developments can affect AO3. We share your concerns. We'd like to share what we've been doing to combat data scraping and what our current policies on the subject of AI are.

Data scraping and AO3 fanworks

We've put in place certain technical measures to hinder large-scale data scraping on AO3, such as rate limiting, and we're constantly monitoring our traffic for signs of abusive data collection. We do not make exceptions for researchers or those wishing to create datasets. However, we don't have a policy against responsible data collection — such as those done by academic researchers, fans backing up works to Wayback Machine or Google's search indexing. Putting systems in place that attempt to block all scraping would be difficult or impossible without also blocking legitimate uses of the site.

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. In many cases, AI data collection traffic relies on the same techniques as the legitimate use cases above.

Once we became aware that data from AO3 was being included in the Common Crawl dataset — which is used to train AI such as ChatGPT — we put code in place in December 2022 requesting Common Crawl not scrape the Archive again.

We cannot go back in time to stop data collection that already occurred, or remove AO3's content from existing datasets, as much as we may dislike that it happened. All we can do is attempt to reduce such collection in the future. The Archive's development team will continue to be on the lookout for individual scrapers collecting AO3 data, and to take action as needed.

Likewise, our Legal committee has and will continue to serve the OTW mission of protecting fanworks from legal challenge and commercial exploitation. This includes their position that users should be allowed to opt out from having their works incorporated into AI training sets, a position that they have presented to the U.S. Copyright Office. They, too, will continue to keep pace with this developing field.

What can I do to avoid data scraping?

You may want to restrict your work to Archive users only. While this will not block every potential scraper, it should provide some protection against large-scale scraping.

AI-generated works and AO3 policies

At the moment, there is nothing in our Terms of Service that prohibits fanworks that are fully or partly generated with AI tools from being posted to the AO3, if they otherwise qualify as fanworks.

Our goals as an organization include maximum inclusivity of fanworks. This means not only the best fanworks, or the most popular fanworks, but all the fanworks that we can preserve. If fans are using AI to generate fanworks, then our current position is that this is also a type of work that is within our mandate to preserve.

Depending on the circumstances, AI-generated works could violate our anti-spam policies (e.g. if a creator posts a significant number in a short time). If you're uncertain whether a work violates our Terms of Service, you may always report it to our Policy & Abuse team using the link at the bottom of any page, and they can investigate.

This statement reflects AO3’s policy at the time of writing, as we wanted to be transparent with our users about what our current stance is and what can be done – and is being done – to mitigate scraping for AI datasets. However, these policies are also under discussion internally among AO3 volunteers. If we agree on changes to these in the future, those will be announced publicly; additionally, if there are any proposed changes to the AO3 Terms of Service, they will be made available for public comment as is required of any and all changes to our Terms of Service.

We hope that this helps to make things more clear – this is a complicated situation, and we’re doing our very best to address it in a way that doesn’t compromise AO3’s principles of maximum fanwork inclusivity or legitimate uses of the site. As discussions and approaches evolve, we will keep our users updated.

Actions

Comments

Sorry, this news post doesn't allow comments.

Pages Navigation

Mr_Understated Mon 15 May 2023 12:24PM UTC

I hope AI Stories don't become the norm. As much as I do enjoy the concept of AI, I do not enjoy people trying to shamelessly replicate humans with AI that don't even know what a finger is or Character that was introduced a paragraph ago. A lot of people using AI are just using it to be unoriginal and copy someone else.
Not to say all AI work is bad, perhaps they could get their own Tag or something. But many people aren't going to use that and just try to claim that a human made it. What an unfortunate circumstance.

Comment Actions
- Thread
Account Deleted Mon 15 May 2023 12:25PM UTC

AI cannot be a ‘fan’ of something, and cannot produce ‘fanwork,’ only plagiarized spam cobbled together from stolen work. The current policy is no different from giving Ao3 users the green light to copy-paste writing by other authors and pass it off as their own. Theft doesn’t suddenly become a victimless crime just because someone invents a machine that obscures the identity of the victims.
Humans are not using AI to create ‘fanwork’ - they are using automated tools to exploit and steal from other humans. You can’t be ‘inclusive’ of this kind of anti-social behavior without allowing someone to be harmed.
The very least Ao3 could do would be to take a stand against allowing the site to be flooded with spam. The alternative is that the Archive will become unusable by human beings looking for human-made works - which is probably where we’re headed, and that’s incredibly unfortunate.

Comment Actions
- Thread
1. orangepurple Mon 15 May 2023 04:54PM UTC
  
  If taking other people's ideas is theft, then aren't all fanfic authors 'stealing'?
  Who is being harmed by the AI? Are fanfic authors losing money from it?
  
  Comment Actions
  - Thread
  - Parent Thread
2. 1. TexasDreamer01 Wed 17 May 2023 02:21AM UTC
    
    If corporate IPs had their way, then yes, fanfic authors would be considered stealing - this is why AO3 doesn't want their users to post links to sites like ko-fi, so they can adequately defend the the works hosted on their site are complying with the Fair Use clause of copyright.
    The companies that own AI are making money off of its use, and ergo off of the data that their bots have been scraping. Provided the myriad lawsuits currently ongoing against many of these companies succeed in their arguments that using the scraped materials constitute copyright infringement, it may or may not result in the source materials themselves (i.e. fanfics) being subject to lawsuits (again). Because of this, allowing AI works on the site could potentially open AO3 - and maybe, though hopefully not, the users - up to valid DMCAs and accompanying copyright lawsuits.
    
    Comment Actions
    - Thread
    - Parent Thread
Sugarysweet Mon 15 May 2023 12:34PM UTC

I’m glad to hear ao3 is putting measures in place to stop ai from exploiting artistic works. Especially since this is fanfiction which we do not get any money from and should never be used to make a profit.

Comment Actions
- Thread
butterflyslinky Mon 15 May 2023 12:38PM UTC

Having skimmed the comments, I have one question about the fears that AI will be used to spam a bunch of low-quality work on the site: Why would anyone do that?
AO3 does not allow commercialization of any work, so it can't be monetized. There's no algorithm to game for popularity. Readers tend to click out of stories that don't meet their standards pretty quickly, and a lot of them are very picky about what they open to begin with. What does anyone have to gain from posting AI-generated fanfic rather than write it themselves?
If someone wants to use AI to create a non-commercial product rather than the joy of creating, that's on them. I don't see it becoming a huge issue for this space because there's nothing to gain from it other than short-term popularity and long-term derision.

Comment Actions
- Thread
1. Kereudio Mon 15 May 2023 03:42PM UTC
  
  Lots of reasons, but the biggest one being: annoying everyone else. There are a lot of people on this site who hate that it hosts what they deem to be problematic content and want to spam and annoy the people/shippers/etc of the things they've deemed to be wrong. I'm in a ship fandom where, despite the tag being relatively slow for the past decade or more, has started to be inundated with low effort, likely untagged AI spam. Unfortunately, when I reported them, I was told that they weren't posting frequently enough to be spammy, and so I had to sit there and deal with it via a site skin until the muting function finally rolled out. Spamming as protest is hardly a new thing, and unfortunately AI has just made that leagues easier.
  
  Comment Actions
  - Thread
  - Parent Thread
2. 1. butterflyslinky Mon 15 May 2023 03:52PM UTC
    
    I understand why that would be frustrating, but banning AI-generated works wouldn't resolve the problem. People are going to be annoying no matter what tools they have at their disposal. And as stated in other comments, there's no way to know for sure if a work is an AI-written troll fic or someone who's genuinely trying and just bad at writing.
    
    Comment Actions
    - Thread
    - Parent Thread
  2. 1. Kereudio Mon 15 May 2023 04:11PM UTC
      
      No, you're right, and I agree that banning AI works could only be symbolic at best and not practical, I didn't say that it would resolve the problem. I was merely answering the question you posed of why someone would post AI generated fic with one reason. :) (Not an aggressive smiley, genuine!)
      
      Comment Actions
      - Thread
      - Parent Thread
konykorynao Mon 15 May 2023 01:27PM UTC

oh

Comment Actions
- Thread
westiec Mon 15 May 2023 02:21PM UTC

I fail to see how you can understand why fans want to avoid having our hard work scraped and fed into a language learning machine and then turn around and say that you welcome the output of those same machines and will treat it as legitimate fanwork. I look forward to a proposed change to the ToS on this subject.

Comment Actions
- Thread
rinny (snowbloom) Mon 15 May 2023 03:12PM UTC

nah not yall not banning AI generated content here 💀 isn’t your aim to preserve all fanworks? i really don’t think a machine, a bot, a pile of code, qualifies as a fan. the person simply inputs directions and generates work, with near to no effort placed into it. ordering an AI to spit out a piece of soulless stuff is definitely NOT comparable to dedicated fans pouring hours of time and mountains of effort into creating heartfelt content.
don’t slap me with that “not prohibited” BS. you can simply add it if you so wished. by claiming that AI generated stuff can be qualified as fanworks, this is an insult to actual content creators.
that said, im glad actions are being made against scraping. kudos to that.

Comment Actions
- Thread
MysteriousSunshine Mon 15 May 2023 03:12PM UTC

I really appreciate the stance on AI-generated fics. While I write all of my fics myself, I recognize that AI is a powerful accessibility tool that may allow someone to engage in fanfiction who might otherwise not have been able to.
As a blind author, I've been able to use AI to help improve my writing. It does a fantastic job at explaining visual elements (clothing, the layout of a city/building, facial expressions, etc.) that I can then encoorperate into my own unique writing style. Before, it took me hours of research just to find out the layout of Queens. Now, I can focus on writing and use AI as a tool.
I think it's natural and crucial to be wary of AI and the potential harm it could bring, but when we make sweeping generalizations about it and the people who use it, that's when we become exclusionary and unfair.

Comment Actions
- Thread
1. thebadwolf Mon 15 May 2023 03:20PM UTC
  
  Yea I think some people (not everyone of course) are forgetting about differently abled people and how they are using AI. I think as long as you are writing your fics and just using the AI as an aid most people wouldn't have a problem with that.
  
  Comment Actions
  - Thread
  - Parent Thread
2. TexasDreamer01 Wed 17 May 2023 02:24AM UTC
  
  I think using AI for research is an excellent use, particularly for establishing descriptions of unknown places, and a good distinction to make from people who use it wholesale to ghostwrite for them.
  
  Comment Actions
  - Thread
  - Parent Thread
glow Mon 15 May 2023 03:52PM UTC

This is not it. Do better.

Comment Actions
- Thread
wednesdaysky Mon 15 May 2023 04:59PM UTC

Every last person who decided it's okay for AI generated "fic" to be posted here should be disgusted with themselves. This place is for fanworks. Machines are not "fans" and they inherently cannot "write" any "fanwork" through regurgitating human fans' plagiarized work. This garbage has no place on the archive.
I hope during the next election we have the opportunity to vote out every board member who thinks this outrageous policy is okay.

Last Edited Mon 15 May 2023 05:00PM UTC

Comment Actions
- Thread
sevenangels Mon 15 May 2023 05:03PM UTC

Wow this brought up a lot of arguments.

Comment Actions
- Thread
Coroniel Mon 15 May 2023 05:25PM UTC

As a person who has never used any of the current AI generation (someone I know did show me its workings one day but that was it), I have zero issues with AO3's stance. It's a difficult situation where imo banning AI works will do little to nothing. I haven't seen any such works so far, but if they pop up in my fandoms I'll read them if they interest me or I'll ignore them, as I do with every other work on this archive. Hopefully the situation doesn't change much from what it used to be.

Comment Actions
- Thread
tenrousei_kuroi Mon 15 May 2023 09:07PM UTC

Okay we need to clear something up here vis a vis "you cannot ban AI content because it would be ineffective" and "trying to do so would be the same as antis asking for problematic content to be banned".
1. Ao3 already has a similar unenforceable ban in place: original fiction. Original works must be "fannish" in nature to be allowed. This is inherently unenforceable. They have, to my knowledge, never removed an original work for not being "fannish" enough. It's something you can't prove. The reason this wording is there is so that the OTW is not promising to do significant legal work on behalf of preserving and fighting for original fiction. Not because "real" books are bad, but because it's *not what the archive was meant for*. So yes, you can post your completely original fiction here, but don't expect the OTW to start sending lawyers to defend the *concept* of original fiction in court; it's not what they're here for.
2. This is fundamentally different than the "asking to ban problematic content" issue. The OTW is prepared to go to bat for the sickest, most twisted and depraved content imaginable. They don't just allow it because "it's unenforceable". They *want* to protect it as a fundamental aspect of expression and the archive. That's why a blanket welcoming stance on all AI-generated works is a bit of an issue, because it implies that the OTW is willing to stick their neck out for everything from a person who uses AI to create an outline that they then expand upon (reasonable, imo) to a completely autonomous program that spits out fifteen legally distinct copies of the same story every day without even being fed a prompt (the future we are quickly heading towards).
3. People who want AI works banned are referring to works in the latter of those two categories, by the way, the ones generated completely by a program. A program that in the very near future will likely not need human intervention to "create" content at all. Possibly not even to post it. And as legal precedent is *everything* in the law world, we need to make sure the way we approach AI right now focuses on human use of it, not the inherent workings of the tech itself. That's why it was so concerning when the OTW's legal chair stated that she believed the actual *process* of an AI "learning" via scraped data was "fair use".
The best holding solution, in my mind at least, is to ban AI-works that have not been altered from their raw format. It is just as unenforceable as the ban on non-fannish original content but serves the same purpose. It would mean the OTW is not planning to advocate legally for autonomous content creating bots the same way they do for human fans.
4. As a bonus, I would also like to add that "AI just isn't as good as humans so don't worry" is a very, very, *very* naïve stance. Like unbelievably naïve. Go ask someone from ten years ago if AI-generated content will ever be as good as it is today and see what they say. Hell, go ask someone from forty years ago if AI-generated content would ever even exist. AI-generated works right now are mostly hot garbage, but give them a few more years.
5. The rest of this is going to come down to a philosophical debate. Some people truly do not see a difference between a computer program "reading" a bunch of books and then writing its own versus a human consuming copyrighted works, feeling inspired and then taking a crack at writing something.
In my opinion the difference is obvious: a human being pulls not just from previous works they've consumed, they draw also from life experience, relationships, their own intrinsic talent and more. An AI program cannot do that. It has no friends, no opinions, no lived experiences. It cannot be motivated by anything. For example, a human may choose to write fanfiction that features more diverse characters because they are *seethingly mad* that the creator never originally included any. An AI is capable of writing the same story with the same diverse cast, possibly even being programmed to do just that...but it does not do so out of any sense of progressive fervor.
The OTW is one of the groups that is working directly with congress to decide what way the law falls on these issues. So people are understandably concerned that in their rush to fight for people who *use* AI to help them write stories today, they may help set up the sort of legal precedent that will *tomorrow* protect the more advanced and more autonomous forms of AI-generated content that are fast on their way.

Last Edited Mon 15 May 2023 09:29PM UTC

Comment Actions
- Thread
1. Kereudio Mon 15 May 2023 09:42PM UTC
  
  I've been one of those people in the comments who's been sighing and resigning myself to the idea that we'll never get a ban because of its unenforcability, because I'm just so defeated after so many attempts to report genuine spam (AI or not) and being told that it isn't. But I always had an inkling in the back of my head that was like "well, AO3 technically bans original fiction, but merely assumes good faith that it's fannish, so what's the difference?" but was unable to articulate that, neither to myself let alone a comment. You put it beautifully, thank you.
  
  Last Edited Mon 15 May 2023 09:42PM UTC
  
  Comment Actions
  - Thread
  - Parent Thread
2. 1. BlackKittens Mon 15 May 2023 10:35PM UTC
    
    The problem here is that original fiction *isn't* banned - at least, not in the way everyone keeps proposing AI works should be.
    Anything that is fannish is allowed on Ao3. This includes fanfics, fanart (although fanart can't be hosted here at this time), podfics, meta, and yes, original fiction. The thing is, Ao3 doesn't take it upon itself to decide what counts as fannish and what doesn't. Instead, they allows the user who uploads said original fiction to decide that whether their work is fannish - and if it's posted here, then they probably have - and won't argue with them on it. Because there are no official rules on what qualifies as "fannish," I can't report an original work under the argument that, *actually,* this isn't fannish! Thus, it won't be taken down.
    That is not what's happening with AI, nor is that what the bulk of people are demanding. The most prevalent argument being thrown around is that AI fics are NOT fannish, and thus shouldn't be allowed on Ao3, period. People have gotten mad that fics they've already reported for being AI haven't been deleted from the site yet. People are demanding Ao3 implement a ban that prevents AI from being posted and will delete it if someone posts it anyway.
    Under the original fiction argument, however, AI would *actually* belong on Ao3 so long as the user posting it decides that it's fannish. Which, if they're posting it, means they probably have. Ao3 doesn't have the time, volunteer numbers, or resources to ask every original work author if they consider every single original work they post fannish. That's why they just assume by default. Therefore, they certainly wouldn't ask about every single AI fic, raw or edited. Especially when everyone will just say "Yes, it is fannish!" to keep it on the site anyway.
    And that will the make the bulk of the people complaining furious. They don't want AI to be considered fannish. They don't want it on the site, period. They want it completely banned.
    So where does that leave us? Ao3 either automatically considers all AI works to be fannish and lets them on the site, or they outright ban it. If they ban it, then that means it has to be deleted when it's posted and reported. But since there's no way to determine if a fic is really AI or not, there's GOING to be instances where AI is unknowingly allowed to stay while fics by real humans are mistakenly deleted.
    Look, I hate AI and don't consider it fannish or transformative myself. But I'd rather it be allowed than let my and other creators' hard work be mistakenly deleted. There really isn't a clear cut solution that will protect everyone *and* ban AI. And I'm not being condescending when I say I'm sorry that it's that way.
    
    Comment Actions
    - Thread
    - Parent Thread
  2. 1. Kereudio Mon 15 May 2023 11:09PM UTC
      
      The petition going around actually does assume good faith, though, and even states that there would be no avenue for reporting anything other than a fic that states, outright, that it's generated solely from AI, and that the result from the prompt has been copy and pasted onto the site with no edits. That petition has over 2000 signatures, so it definitely isn't everyone that's voicing their concerns saying otherwise, and a lot of people in this comments section, when pressed, come to the same conclusion: assume good faith, and only remove work that very clearly and explicitly violates that good faith.
      But what a lot of people are saying is that it doesn't matter if the ban actually results in AI fic getting removed, just that it makes it clear that the OTW does not a) support AI fic and b) won't use donation drive money to defend AI works in court. What started this whole controversy was the legal chair, who is from what I understand not an elected position, saying that she agrees with companies like Stability AI saying that their programs and prior scraping are protected by fair use, which puts her at odds with the artists and companies suing stability AI and similar companies and rightfully scares those of us who donated to the OTW, because as you can tell, a lot of us don't agree that it is fair use.
      But also, even though original work isn't banned to the point of removal, and would likely never be removed because of the good faith assumption, it comes with it two caveats: that the OTW will not defend its copyright status in court, because it is considered fannish by being hosted on the archive, and that the OTW will not consider any remixes, fanfics of, etc to be infringing, because that'd be all fair use. Basically, by hosting original work on the archive, you relinquish any right you may have to control whether or not it can be transformed, unlike with published and copyrighted works. It's also bound by the same rules regarding commercialization: you can't, say, link to the Amazon store page to buy the book, as that'd still break TOS.
      
      Comment Actions
      - Thread
      - Parent Thread
    2. 1. BlackKittens Tue 16 May 2023 12:15AM UTC
        
        I haven't read the petition you brought up (I know it's going around, but I haven't read it) so I can't speak on that; however, I have read plenty of comments here and on Twitter that do want a ban, so that's what I based my previous comment on. And I do believe that even if things went your way, they would still scream in fury at the lack of a total ban.
        As for removing fic that openly states it's AI (wholly or not, edited or raw, whatever the Archive would end up going with), I don't see that working either. People would still continue to post AI, as many have said, they would just never admit it's AI.
        I'm sorry, but I just don't understand how a ban on certain content is acceptable when said content is still allowed *so long as it never says it's name.* I get the sentiment people want - that Ao3 stands against AI on principle - but doesn't that principle fall apart when AI ends up utilized, posted, and allowed to stay anyway? I know text can be hard to read on the internet, so again, I'm not trying to be condescending. I'm just genuinely confused on how this can be enforced in a way that isn't hypocritical or contradictory. It seems pointless to me.
        You also say you don't want Ao3 to support AI or donations going to defend AI in court. I understand that, I do. When I first heard about all this, my gut reaction was, "Hell no, Ao3 shouldn't advocate for or defend AI It's art theft!" But to my understanding *now,* that's not what's happening, and people are at best misinterpreting Betsy's words and misunderstanding the situation, or at worst taking everything out of context. And hey, I could be wrong! I'm not an expert on legalities or AI, so I'm not going to pretend I understand everything she said or what happened in court or in front of Congress.
        But from my understanding, the legal issues surrounding AI are a complicated pretzel that go far, far beyond whether the resulting work was created by a person or computer. It's tangled up in a lot of areas that fanworks are ALSO tied up in, so it's difficult to argue one way or another about AI without also implicating fanworks made by humans. I don't *like* that argument, but if that's true, then I don't want be dragged down with AI in terms of being outlawed.
        As for Betsy, she just seemed to be acknowledging that scraping already happened and was trying to find a bright side in it all - she never said AI is hands down on the same level as human made fic, or that she encourages everyone to use it. Beyond that, it read to me like she was trying to defend fanworks in general, which unfortunately AI is tangled up with now.
        And in regards to donations in particular, I want to put a common myth I keep seeing spread to bed: Ao3 isn't using donations to promote or defend corporate interests, including that of AI. When Ao3 uses donations for legal matters, it's to pay the mandatory court and filing fees that come with *going* to court. The lawyers aren't paid because all of them - including Betsy - are volunteers. Ao3 doesn't go to court over just anything, nor do I believe they can.
        They can't just go to court to argue, say, that bread has too much processed sugar in it to truly be bread, so we shouldn't put so much sugar in bread in the factories! ...Okay, weird example, I know, but my point is I don't think Ao3 has, or will ever, randomly go to court to defend AI just because it wants to. If it has, or will, it has be already be related to the stated mission of the non-profit; it's mission to protect and preserve fan works. Regardless of whether we, as fans, consider AI fanworks (*I dont*), if it's being tied up with us in areas that also affect us, well, then Ao3 needs to step in, don't they?
        Again, this is all just my understanding and I'm not going to pretend I know everything for a fact. I don't. I could be wrong on several points here. But I think a lot of people are jumping the gun on this and don't fully understand the implications of what they're asking for either.
        Lastly, I'm confused by your last paragraph. I apologize if I'm misunderstanding you, so please correct me if I'm wrong. But Ao3 isn't arguing the copyright status of original work - US law already defines original fiction, whether it's fannish or not, as automatically being copyrighted the moment it's made. Fanfiction is also copyrighted. If I write a fanfic, I don't own the use of, say the Star Wars universe or any of its characters I wrote in, but I DO own the collection of words I wrote, so Disney can't just take my hypothetical fic for themselves and make a movie on it without my permission. You don't lose your copyright on anything, original or fanfic, by posting here.
        The monetization issue is a related, but wholly different beast. Ao3 is owned by the OTW, which is a non-profit. *Some* non-profits make commmercial income and profit, yes, because there's different types of legal, non-profit organizations, but the OTW is not that type of non-profit. They're the type that's not legally allowed to. They opted to be so because companies like Disney might be more inclined to sue that fannish works *in of themselves* are copyright infringement due to their montization. Ie, Disney loses money that "rightfully" belonged to them when UserxNamex wrote a Stars Wars fic for profit without their involvement. That doesn't mean by posting to Ao3, for free or profit, you give copyright. You actually can't willingly give up copyright at all under US law.
        I'm sorry if I'm not coming across well either.
        
        Comment Actions
        
        Thread
        
        Parent Thread
      2. Kereudio Tue 16 May 2023 03:47AM UTC
        
        I have pretty severe unmedicated ADHD, so I definitely can't respond to all of this, but to the last point: what I was pointing out is that, say, if you host your original book on the archive, AO3 won't defend you if you go to them and say "xxUserNamexx wrote fanfiction featuring my original characters, this infringes my copyright and I want it gone!" ala Anne Rice. They'll say that it's fair use and send you on your way. But, similarly, if you host your original book on the archive, they'll assume that it's made in a fanfiction context, and therefore defend it as such in court, should it come down to that. They won't put any of their resources towards defending it as an original work, such as arguing for the timeline to be extended before it enters the public domain (if AO3 exists that long, just an example off the top of my head.)
        I'm still wrestling with myself my feelings on the whole matter, so I really can't answer your questions. Personally I've resigned myself to this whole AI thing. It's already stolen most of my things I've posted everywhere on the internet and old AO3 accounts/orphaned works, it's already been used to troll my ship communities, and it's already cost me 3 copywriting jobs. I'm pretty broken at this rate, so I don't frankly care too much what the archive chooses to do, because I've already resigned myself to the fact that it's pretty hopeless and yada yada. Not to get depressing, sorry :P I wish I could flip a switch and eliminate AI from existence, but alas.
        
        Comment Actions
        
        Thread
        
        Parent Thread
        
        tenrousei_kuroi Tue 16 May 2023 11:37PM UTC
        
        Ngl I find it strange that this convo thread is going on beneath *my* comment...when mine was the comment specifically saying that it would be good for Ao3 to not openly welcome 100 percent computer-generated content without actually trying to enforce a practical ban, and why the unenforceability is not the point...only to be hit back with an argument that just says some sort of "delete things that are AI" policy wouldn't be enforceable...lol.
        I do not believe that doing so, when worded correctly, would result in people "not tagging AI stuff", either. After all, people still tag original fiction, don't they? Ao3 even has commonly marked tags and categories for it. Hell, most users don't even know truly original works aren't meant to be hosted here. I wasn't expecting some big "AI WORKS BANNED" banner to scroll across the site, moreso a slight alteration to the wording of Ao3's ACCEPTANCE of AI-generated works.
        Perhaps I should not have used the word "ban" at all, as it seems to be confusing people. So let me reword my previous statement to:
        "The best holding solution, in my mind at least, is for Ao3 to welcome AI-generated works that have been altered from their raw format, as a form of fan-created work."
        I was going to explain further the fannish original works policy, but I think you've already pretty much covered it here with the extending copyright example so thank you for that. I never meant to imply that the "concept" of original works was under attack or anything as has been bafflingly interpreted by someone a few comments up...Just that there are separate legal issues that come up in the context of non-transformative works that the OTW just isn't committed to advocating for. I didn't realize that was such a confusing point >.<
        I also never meant to imply that the OTW was hiring some shady cabal of lawyers to work for OpenAI or whatever, just that a portion of their donations cover legal fees that may indirectly be working against human writers. I'm not sure if this was unclear or not but I was trying to communicate that organizations like the OTW (who are currently communicating with US lawmakers about the legality and copyright-ability of AI works) should be *careful* in how they define these things. Ie: maintaining a human element lest we accidentally set legal precedent for 100 percent autonomous content creation being protected the way human works are.
        I never, never, *never* would even dream of anyone from the OTW saying "I think this no-humans-involved spam content is just like a fanwork", but there is a concern that imprecisely worded defense of AI-generated fics *directed and posted by humans* may accidentally set groundwork for imbuing completely autonomous content with too many protections.
        Right now things are leaning towards AI-generated works *without substantial human authorship* NOT being copyrighted, automatically or otherwise. If the OTW *OPENLY* welcomes all AI content here, even that which is not currently copyrightable in the US, they are at the very least complicating things for themselves in terms of whether or not they would defend such works and in what capacity.
        At the very least, users of the archive should not be confused if they post a completely AI-generated work, it gets stolen and resold elsewhere and then suddenly the OTW (which promises to protect fan communities from corporate exploitation, as stated in the very post) suddenly *isn't* helping them.
        (On an unrelated note, I totally get the copywriting gig trauma. A company I used to work for as an editor recently switched completely over to AI and layed off about half their staff and I'm like, wonderful short sightedness, guys. Look forward to seeing you desperately try to rehire when your clients suddenly realize they hold no copyright over the content you've "sold" them lol)
        
        Last Edited Tue 16 May 2023 11:51PM UTC
        
        Comment Actions
        
        Thread
        
        Parent Thread
        
        Kereudio Wed 17 May 2023 12:08AM UTC
        
        Sorry if your inbox got spammed, aha. I completely forgot at some point in the night what comment thread I was in at any given time. :) FWIW, I agree with everything you said, and thought you were quite clear, and have spent the last few comments trying my best to restate the point in my own words both for my own and others understanding (and also come to terms with my own feelings on the subject.)
        I especially agree with that I don't think a restriction of/ban on/whatever you wish to call it of AI works would decrease the amount it gets tagged... Both for the reasons you said, people perhaps not realizing, but also because I think the majority of people on AO3 respect the sanctity of the human artist, or at least understand it. The few people in this comments section who have admitted to using AI have said they tag it properly, and as others have said elsewhere, there's no profit to be made or algorithmic clout to be gained by posting low effort, mass produced AI content, so I would assume most people posting it would have some modicum of effort put into refining and editing it.
        Of course, as has been my experience on a few (thankfully few) occasions, there are always people who just... want to troll a tag they don't like, and I am worried that this open doors policy will encourage them to do that more. I sincerely hope that won't be the case, but.
        
        Last Edited Wed 17 May 2023 12:08AM UTC
        
        Comment Actions
        
        Thread
        
        Parent Thread
        
        tenrousei_kuroi Wed 17 May 2023 01:01AM UTC
        
        Yeah, I agree lol. I am less worried about the idea of individual people mass spamming Ao3 tags *right now* (because they can currently do that just as effectively by typing out a billion two sentence microfics, which takes actually *less* time than prompting an AI and then copy-pasting the results as a super short 'fic' is probably no greater in wordcount than an AI prompt would be) and more concerned about the potential for future programs to mass-produce "literary" content more or less autonomously, and likely on sites where the person behind it can make some money.
        And while the OTW isn't in any way the deciding authority on any of this, they are definitely a player. I just don't want to see the day where a computer-generated work of fiction is copyrightable and while that's currently not the case, making wording mistakes like saying "we defend AI-generated works" instead of "we defend AI-generated works that have been meaningfully altered by a human" is exactly how you accidentally pave the way to change that.
        I think people would appreciate a clearer stance on the OTW supporting *specifically* human-created (even if AI-assisted) works rather than this weirdly vague "all AI stuff".
        Honestly, that probably *is* their stance. But they have not done a good job of stating that lately, which is how this whole kerfuffle started lol. And then moves like just deleting the link to the Betsy interview rather than fully addressing people openly about it is (unintentionally) shady. It's like, is it that hard to say that you consider something made by a person *using* AI a real fanwork but not something done completely by a program?
        So much of these arguments are coming from just...the most bizarre game of telephone. Person A says "ban all AI content", clearly in their mind referring to copy-n-paste words spat out by a program and Person B screams it's not fair because they once had Grammarly rewrite a sentence and they're convinced your gunning for them, lol. Though I'll admit there are two separate issues going on here and the idea that all AI-generated works are inherently plagiarism due to the nature of how LLM's are trained is an adjacent if separate one. I think it's highly unlikely we'll win that battle anywhere or that the OTW would agree, which is why I focused instead on the outcome, the idea of whether or not a person is consistently guiding the AI and editing output or if the program is just chugging along on its own.
        
        Last Edited Wed 17 May 2023 01:04AM UTC
        
        Comment Actions
        
        Thread
        
        Parent Thread
        
        tenrousei_kuroi Wed 17 May 2023 01:14AM UTC
        
        It's likely that they may feel that saying "if fans are using AI to generate works" is clear enough on this matter. I would argue that is simply isn't, as I feel that wording would easily include the idea of someone simply "turning on" a program that then spits out forty-five stories a day. Again, not how AI works are typically generated *now*, but I'd say very likely to be the case in about...oh, five minutes or so given how quickly the tech is evolving. And it is not currently in line with how copyright law is talking about AI-generated works, which (*crosses fingers*) seems to be sticking to its guns with requiring "meaningful" human alteration. So I feel like a huge disconnect is happening.
        If there comes a day when there is more AI-generated stuff on the archive than human-created works then it will essentially just be a site *for* that type of content. And like I've said before, if someone's only argument against that being the future is "AI-generated stuff isn't that good so people won't want to read or post much of it", then that person is either 12 years old or an adult who has been living under a rock their whole life. People consume "soulless", corporate, design-by-committee art all the time and I don't think it will be that long before an AI-generated piece is indistinguishable from "bad" human art.
        
        Comment Actions
        
        Thread
        
        Parent Thread
3. thebadwolf Mon 15 May 2023 11:11PM UTC
  
  "The best holding solution, in my mind at least, is to ban AI-works that have not been altered from their raw format."
  Yes I actually agree with that. If you are just copy and pasting what an AI puts out then there is a problem there. If pop out 100 words to just kinda of get started or get your head around what you're writing and then turn it into a 5000 word masterpiece on your own, I don't see anything wrong if that. Or if you make outlines. Like with chatGPT I can describe a story that I have planned out and I ask it to make an outline based on that information. All it is doing is helping me put my information in some kind of order/layout that makes sense.
  I don't think people like me who rely on AI in some form (mostly for editing) should be worried. Lots of people who are differently abled do need this tech to help them write. And there's nothing wrong with that.
  
  Comment Actions
  - Thread
  - Parent Thread
4. MintGreen_and_NeonPink Tue 16 May 2023 02:14PM UTC
  
  Wonderfully said, thank you
  
  Comment Actions
  - Thread
  - Parent Thread
5. likegoldandfaceted Sat 20 May 2023 05:53AM UTC
  
  Original work is entirely allowed. I'm not sure where you came to that conclusion; it's even a properly wrangled tag.
  
  Comment Actions
  - Thread
  - Parent Thread
6. 1. tenrousei_kuroi Sat 20 May 2023 12:53PM UTC
    
    From reading the TOS where it clearly states that Ao3 openly welcomes original works that are “fannish” in nature but not completely original works “in order to avoid becoming a general repository for all sorts of creative works.” Works uploaded under the original works tag are done so on the honor system and by using the tag you are declaring that you feel your works fall under the context of being “created in a fannish context.” Ao3 and the OTW will protect them as such, but not as a non-fannish work.
    
    Comment Actions
    - Thread
    - Parent Thread
Answrs Mon 15 May 2023 10:31PM UTC

"we're protecting users against data scraping for ai generation!"
great!
"also all content made out of unconsented scraped fanfic is great and we encourage people to post it here."
fucking yikes.

Comment Actions
- Thread
Trudemaethien Mon 15 May 2023 10:44PM UTC

writers should be able to protect and restrict their writing—more than just “maybe”—from being scraped. The thing about the scraping is that it is taking fanmade, human-made fanworks and using them for ultimately commercial purposes. sure, the AI can be used to create more fanwork, but we all know it’s not limiting itself to doing only that.
I want it kept as far away from AO3 as possible, and if people want to practice with it, fine, do it elsewhere. But the AI is not the fan. The AI is doing the work, the fan is not. There is a disconnect there, so I say what the AI spits out is not actually a FANwork. It’s an AI work.

Comment Actions
- Thread
kissingkeith Mon 15 May 2023 11:19PM UTC

Preserving computer generated plagiarism should not fall within the Archive's mandate.

Comment Actions
- Thread
1. ofthesummer Wed 17 May 2023 01:19AM UTC
  
  We need to get rid of AI-lover Betsy Rosenblatt. She's the board member who wants the site to be flooded with AI fics and who has spoken publicly about how great AI is at every public opportunity.
  
  Comment Actions
  - Thread
  - Parent Thread
SailorStarDust1 Mon 15 May 2023 11:24PM UTC

When that (AI?) Bot Scraping situation in numerous fanfic comments happened in January-February (my months might be slightly off but you get the point), I initially shrugged things off and marked them for spam...But receiving 8 spam scraping comments en masse led me to finally restricting my account.
As a hobbyist writer since 2003 and in light of those recent events, I can't help but feel frustrated by this new decision. I strongly urge Ao3 to instead implement—at the very least, to meet everybody halfway—an "AI Generated" tag so writers can pick whichever suits them. Edit: A word.

Last Edited Tue 16 May 2023 12:23AM UTC

Comment Actions
- Thread
1. vanyamire Tue 16 May 2023 02:54AM UTC
  
  A tag would be tacit endorsement though. I do not think a tag is a good enough solution, not when that would mean AO3's legal and monetary resources are going towards defending AI crap alongside our own work.
  
  Comment Actions
  - Thread
  - Parent Thread
2. 1. FireForEffect Tue 16 May 2023 04:14AM UTC
    
    If AI-reworked existing written content counts as transformative work it accords machines the artistic rights and considerations we've previously accorded humans. For what?
    
    Comment Actions
    - Thread
    - Parent Thread
  2. SailorStarDust1 Tue 16 May 2023 02:35PM UTC
    
    I feel uncertain what else can be done if the Board themselves are seemingly "for" AI, you know? *Sighs* While AI tools can be a beneficial thing (video game design, for instance) the increasing prominence of theft/plagiarism even within fanworks (whether fanart, or for instance the fanfic comment scraping) leaves me incredibly frustrated how it's seemingly so easily...acceptable.
    
    Comment Actions
    - Thread
    - Parent Thread
  3. 1. vanyamire Tue 16 May 2023 02:49PM UTC
      
      I understand :( but we gotta keep pushing them. I mean, it's our archive too. Like, I just learned that the chair of their content committee is Betsy Rosenblatt as well. So we've got to ask them if that's the reason they're being so pro-ai.
      Enforcement of policy against ai works is one thing. I'm concerned that this behaviour could carry forward during court or legal proceedings. I mean, look at what they've already said to the US copyright office. They asked for opt out- an ineffective tool that places all the burden on the exploited writer/artist. I just expected a much firmer stance from the OTW when they've said multiple times they'd protect us from commercial exploitation.
      I just feel like we shouldn't give ground under these circumstances. Anyway, I understand your frustration all too well T_T
      
      Comment Actions
      - Thread
      - Parent Thread
    2. 1. SailorStarDust1 Tue 16 May 2023 03:17PM UTC
        
        Ugh. I just don't understand their thought process.
        An anecdote: Last November(?) during board elections, there was a situation where somebody pro-censorship was to be voted in, but thankfully the community pushed back, hard. I thought to myself I'd be sure to donate this year to prevent that from happening again...But now, this entire AI situation has left an incredibly bad taste in my mouth. Killed any recent writing enthusiasm on my end.
        I believe all we can *realistically* do is continue expressing dissatisfaction to the Board themselves, here and through their contact info: https://www.transformativeworks.org/contact_us/
        
        Comment Actions
        
        Thread
        
        Parent Thread
In_Which_Magic_Is_Real Mon 15 May 2023 11:41PM UTC

There are many comments how fics written by AI should not be considered as fannish, but they all seem to gloss over one important part; the fan who prompted the AI, and uploaded it to Ao3.
The legality of using AI aside ( and I know this is a huge thing to put aside), one of many question to ask is if it alright to dictate how a fan chooses to show their passion for their fandoms. If a fan, for whatever reason, chooses to use AI to write, is it up to Ao3 to decide whether it is an appropriate? Is it up to the other users? Or is it up to the fan themselves? Who, in the end, has the final say about that?
Policing how a fan chooses to create is more alarming to me then the fan using AI. As long as it is not illegal, which AI is currently not, Ao3 as a rule tries not to put limits on how a fan expresses themselves.
I think this current stance by Ao3 on not banning AI because a fan is involved is probably the best compromise right now. I think the legal aspect is much more important, with much more far reaching consequences then arguing about how much AI is or is not able of being used for fandoms.

Comment Actions
- Thread
1. Sherloqued Tue 16 May 2023 12:41AM UTC
  
  How do we know that it is strictly 'fans' using the scraped material though? And not some other entity looking to capitalize on someone else's work?
  
  Comment Actions
  - Thread
  - Parent Thread
2. 1. butterflyslinky Tue 16 May 2023 01:23AM UTC
    
    People who want to capitalize on other people's fanfics don't need AI to do that, they just outright steal and resell it. And since selling fanfiction is legally dubious to begin with, that's easy to shut down. Same thing with AI--even if someone is using the scraped data to write more fanfiction, they aren't getting any real benefit from it.
    AO3 doesn't have any sort of monetization, so the most someone could do with fic generated from scraped data is be mildly annoying with spam.
    
    Comment Actions
    - Thread
    - Parent Thread
  2. 1. Sherloqued Tue 16 May 2023 01:58AM UTC
      
      I sure hope not, because the ultimate irony of ironies would be if a person's work from their heart and soul became the protected, inaccessible intellectual property/trade secret of someone else or a company!
      
      Comment Actions
      - Thread
      - Parent Thread
3. Kereudio Tue 16 May 2023 03:12AM UTC
  
  AO3 has a mission statement to be maximally inclusive and as non-discriminatory as possible as to what constitutes a valid way to engage with fandom, but that doesn't actually make them the arbiters of what is and isn't acceptable. They aren't the only website where one can engage in fandom activities, and other websites don't disallow AI fic. Furthermore, their code is open source, so if someone decides that AI fic is a valid way to engage with fandom, they're more than welcome to the software, they just have to provide the server.
  It is called Archive of *Our* Own, and if the majority of members eligible to vote, and the majority of volunteers, decide that AI fic isn't welcome, then that's the decision they're allowed to come to, and is what represents the wants of the community here. It doesn't mean that they've handed down an iron rule that everyone, everywhere, must follow, just that that's what they've decided is or isn't acceptable on this singular platform.
  
  Comment Actions
  - Thread
  - Parent Thread
4. 1. In_Which_Magic_Is_Real Tue 16 May 2023 01:13PM UTC
    
    The problem here is that there is a conflict between banning content people don't want on this plateform, and the primary purpose of Ao3.
    Let's say every single Ao3 user voted, and 51% voted to ban AI. Does that then give that 51% percent the right to make the changes unilaterally, even though 49% were against it? Especially since the reason Ao3 was created in the first place was as a space where fans didn't have to fear their work being removed even if a majority wanted them gone.
    In my mind, replace AI with any other subject and the obvious answer is no, banning is not ok. But in the context of AI, where people feel so strongly about if AI can even be considered fan work, the answer becomes murky, and I worry that the anger is clouding the very real and very scary impacts of opening a way of banning on this plateform based on the wants of majority.
    It is more than ok to not interact with AI generated content, but the burden should not be on Ao3 to remove it. Ao3 has always been about moderating your experience so as to avoid seeing content you don't want, not removing that content
    
    Comment Actions
    - Thread
    - Parent Thread
  2. 1. Kereudio Tue 16 May 2023 03:13PM UTC
      
      I think that's a logical fallacy. There is no logical path that goes from "declaring AI generated work as plagiarism and banning it under the current rules against plagiarism" to "no incest fic." AO3 has, time and time again, gone to bat for things like problematic content and other objectionable - *fictional* - content, regardless of what anyone says. And the loudest opponents to that aren't OTW members, so they don't get a say in what makes it into the policy.
      AI isn't "problematic content", and the debate lies with whether or not it counts as plagiarized content, which is already banned by the tos. If someone copy and pastes your fanfic word for word but changes the names, they don't tell you to curate your experience and ignore it, they take it down. This isn't about whether or not AO3 should back out of it's policies to never ban content unless it's plagiarized or a straight up violent threat or not a fic at all, it's about whether or not a relatively new thing that wasn't even considered when the TOS was written should be folded into existing definitions of plagiarism. Every educational institution has done so, so it's not as though the AO3 doing the same would be completely without merit.
      
      Comment Actions
      - Thread
      - Parent Thread
    2. 1. stormyseasons Wed 17 May 2023 11:46AM UTC
        
        Yes.
        And... I seem to remember hearing that some of the commercial publishers, writing magazines, (Amazon's kindle ebook program?) have also refused to take 'AI-generated' work too... what was the reason they cited, was it plagiarism issues, uncertain copyright legality issues, or something to that effect...
        
        Comment Actions
        
        Thread
        
        Parent Thread
bluevinegar Mon 15 May 2023 11:45PM UTC

What a difficult situation... I'll admit, I don't especially understand the purpose of "AI" generated fanfiction. Writing fic is a particularly indulgent activity for me, so I don't quite grasp the point in having a machine take over. Especially on AO3, where you cannot really earn money or build clout for quickly churning that kind of content out anyway.
Regardless, thank you all for your transparency. Though I'd rather not have it present on the archive, I appreciate the steps taken to prevent scraping and understand the difficulties that would arise in a blanket ban. (Like others have stated: the concern of drawing a line for other machine-supported writing, of innocent works being flagged, people simply just lying/omitting tags and posting it anyway, etc.)

Last Edited Mon 15 May 2023 11:46PM UTC

Comment Actions
- Thread
hiromiroll Mon 15 May 2023 11:58PM UTC

after reading all these comments, i feel like we can agree that it is archive of *our* own, and not archive of AI's own.
i dont think that AI is a bad thing, i just think that the AO3 is just not the place for it.
that's just my opinion though.

Last Edited Tue 16 May 2023 12:06AM UTC

Comment Actions
- Thread

AO3 News