New Rule Regarding Generative AI

[#1 Taln Fan he/him]

Hey all, I want to let you know that we've updated our Forum Policies with some new rules regarding posting and discussing AI generated content:

Quote

AI-generated or assisted content is not allowed anywhere on the forums or discord. Sharing or discussing AI prompts ("prompt engineering") is similarly not allowed. Discussing the technology itself, or media created with it, is permitted.

If you suspect that something someone posted is AI-generated, we ask that you simply report the post rather than confronting the poster directly, and the staff team will deal with the issue as needed. 
In addition, we've also clarified our stance on how the rep system is to be used:

Quote

Reputation points manipulation is prohibited. This includes asking for rep, telling others to rep/upvote someone else, and spam-repping someone to get them to a certain level or just for the sake of upvoting them rather than because you like each individual contribution. All parties involved may have their ability to use the reputation system suspended or revoked.

If you have any questions or concerns, feel free to either reply in this thread, or PM me or another moderator. 

[Frustration]

I'm not sure I entirely understand the reason for the AI rule. There haven't been a lot of AI related posts so I don't think it has generally been a problem. Is this just to get ahead of a potential issue or is there something else?

[Chaos he/him]

We would like our policies to be relatively consistent with Discord, and there have been some AI related content posted. This simply clarifies that people can't just put out AI generated stuff; it will be removed. It's come up some and will only become more prevalent. We do not need to wait for it to become problematic to act. 

[Frustration]

I wasn't trying to say that it had to be an issue first, I just was wondering as to the reasoning behind it.

Though I suppose that answers my question, so thanks!

[Duxredux he/him]

A couple questions for clarification. Where do profile pics and cover photos fit in on here and Discord? I would guess this would be the most common source of AI generated content. Do I need to draw my own top hat wearing duck and replace this one from 2023?

I will admit to having AI check my logic when someone asks really esoteric questions. Last one was about an Attractor Fabrial diving bell and I wanted to get the physics right for the mechanics behind SCUBA and the physiological constraints on a human diver for my response - or at least be told if I was missing something fundamental. Would this constitute AI assisted content, even if all of the text was personally written?

Like Frustration, I do not yet fully understand the rationale and I want to be in compliance.

[Chaos he/him]

3 hours ago, Frustration said:

I wasn't trying to say that it had to be an issue first, I just was wondering as to the reasoning behind it.

Though I suppose that answers my question, so thanks!

Oh, no worries. I just thought you were meaning, "There hasn't been a problem, so why do it?" hence my response. 

3 hours ago, Duxredux said:

A couple questions for clarification. Where do profile pics and cover photos fit in on here and Discord? I would guess this would be the most common source of AI generated content. Do I need to draw my own top hat wearing duck and replace this one from 2023?

I will admit to having AI check my logic when someone asks really esoteric questions. Last one was about an Attractor Fabrial diving bell and I wanted to get the physics right for the mechanics behind SCUBA and the physiological constraints on a human diver for my response - or at least be told if I was missing something fundamental. Would this constitute AI assisted content, even if all of the text was personally written?

Like Frustration, I do not yet fully understand the rationale and I want to be in compliance.

AI generated profile pics / cover photos would indeed not be allowed.

For your second thing, the text is human made; how you research is your own business. (But, pasting an AI generated summary for someone else would be not allowed, for example.)

[Through the Living Shadow he/him]

Hhmmmm

 

im glad I saw this

Cuz I know I was discussing some stuff that AI posted

: )

I shan’t do it again!

[Through the Living Hope]

On 4/6/2026 at 7:57 PM, Chaos said:

Oh, no worries. I just thought you were meaning, "There hasn't been a problem, so why do it?" hence my response. 

AI generated profile pics / cover photos would indeed not be allowed.

For your second thing, the text is human made; how you research is your own business. (But, pasting an AI generated summary for someone else would be not allowed, for example.)

Should past posts with AI images be edited to remove them if possible? Or is this just moving forward?

[Chaos he/him]

2 hours ago, Through the Living Hope said:

Should past posts with AI images be edited to remove them if possible? Or is this just moving forward?

Don't stress about past posts with AI images; that would be silly for someone to search through their posts to find, and impractical to police even if we wanted to. 

[Immortal Platypus]

can I ask why prompt engineering is banned? just to be consistent with discord? is there a specific reason it's banned there? And where does it cross the line from prompt engineering to discussing the technology itself?

[Qianweilian He/him]

On 4/5/2026 at 3:56 PM, #1 Taln Fan said:

If you suspect that something someone posted is AI-generated, we ask that you simply report the post rather than confronting the poster directly, and the staff team will deal with the issue as needed. 

How will you determine how a post is AI? Because, while it is obvious for many posts, there are many situations where it is impossible to determine origin. How stringent are the mods planning on being regarding this new rule.

Also, will there be an appeal process different from regular staff decisions? Because this rule, depending on its enforcement, is likely to be subject to a staff member's best judgment, and not any sort of clear cut methodology.

[Frustration]

Adding to this we have several non-native English speakers on the shard. Are they forbidden from using AI translations, or would that use be permissible?

Edited April 8 by Frustration

[Chaos he/him]

8 hours ago, Immortal Platypus said:

can I ask why prompt engineering is banned? just to be consistent with discord? is there a specific reason it's banned there? And where does it cross the line from prompt engineering to discussing the technology itself?

For example. We disallow AI fanworks in this, yes? We don't want to allow discussion of "This is how you can get the best Jasnah Kholin generated". So it's things like that. I don't think those conversations happen often, but that's covering our bases.

6 hours ago, Qianweilian said:

How will you determine how a post is AI? Because, while it is obvious for many posts, there are many situations where it is impossible to determine origin. How stringent are the mods planning on being regarding this new rule.

Also, will there be an appeal process different from regular staff decisions? Because this rule, depending on its enforcement, is likely to be subject to a staff member's best judgment, and not any sort of clear cut methodology.

Yes, it will indeed require judgment. We don't have much interest on doing extensive research or investigation in particular posts to discern what is AI generated. I imagine we will give plenty of grace and benefit of the doubt, and I imagine most people will just be, "Okay, cool, don't post stuff from ChatGPT here." 

If you are worried about using em dashes, worry not. We love our em dashes here. 

So, yeah, particularly as the tech advances, this will be harder and harder to discern. I would rather be in a place with strong policies than just say, "Eh, anything goes." I don't particularly want to be on a forum where no one is writing their own text and it's all just AI generated topics with AI generated posts. We want people to communicate with each other. 

No plans currently for any special appeal process. We don't have a formal appeal process anyway. Let a mod know if there's a false positive. I don't think we have removed anything on the forums as of yet, so I'm not super concerned right now. We are open to adjustments as things progress.

If anyone has had their paper they worked hard on be accidentally flagged as AI, I would not worry. 

4 hours ago, Frustration said:

Adding to this we have several non-native English speakers on the shard. Are they forbidden from using AI translations, or would that use be permissible?

I would not consider that generative AI. That's been around for a long time. 

Now... If someone's not writing anything and are just having an LLM generate them a whole topic, that would be different matter. 

[NerdSandwich she/her]

18 hours ago, Chaos said:

em dashes

DID SOMEONE SAY EM DASHES?!!!

Thanks Chaos that is very sensible (Uhh not sure if that's comforting from me considering how young I am...)

Anyways, I don't have access to AI but I am wondering:

What if someone brainstorms using it or smth? Where do you draw the line?

and

What if someone doesn't know about the rule? Will they be punished severely?

 

 

[AonEne he/him]

47 minutes ago, NerdSandwich said:

What if someone brainstorms using it or smth? Where do you draw the line?

and

What if someone doesn't know about the rule? Will they be punished severely?

Generally, if all of what they're posting is their own writing and thoughts, we're not going to moderate someone for having tossed ideas back and forth or researched with an AI. We don't really want people to post something that's just "I was brainstorming with AI and it said this, what do you think?" and that's all - we want people to talk about the things they think - but if AI suggests an idea to you and you start talking about your thoughts and expanding on that idea, that should be okay. 

No. People should be reading the rules when they join but some might have previously done that and then missed that the rules got updated. It's a minor violation more like double posting or sending one-word/emoji posts or something, we'll let the person know it's against the rules and remove the content in question, and it's only if they keep doing it repeatedly despite having been personally told not to that we'll moderate more severely. 

[NerdSandwich she/her]

Ok thanks!

[Chaos he/him]

57 minutes ago, NerdSandwich said:

DID SOMEONE SAY EM DASHES?!!!

Thanks Chaos that is very sensible (Uhh not sure if that's comforting from me considering how young I am...)

Anyways, I don't have access to AI but I am wondering:

What if someone brainstorms using it or smth? Where do you draw the line?

and

What if someone doesn't know about the rule? Will they be punished severely?

As I said before, how people research is their own business. If the text is written by you, cool.

Expanding off what Ene said: this can apply to essentially any rule. We are happy to answer any questions, but hypotheticals like this can be very challenging to answer. Context heavily matters. Most situations, as Ene said, I think it'd be a minor thing where once a person is told, they go, "Okay cool no worries I won't continue." But I could imagine scenarios where people are being especially egregious which could lead to harsher punishments, for example, imagining someone who is posting a bunch of art, they say they made it, they repeatedly affirm they created it, and it comes out it was AI generated. Like... there's a level of deception there that violates the spirit of our Code of Conduct as well as this rule that in my opinion would warrant further action.

But, ultimately, it's very hard to say. There has been nothing beyond "Someone shared a meme that looked AI, and we removed it in accordance with the policy". We are going to be discovering edge cases and making determinations. Maybe there will be a major issue or event that leads to something, but it is highly dependent on context of the situation, so I can't possibly know how we would rule.

However, I would say that we always make decisions as a group. No one person is making any determination. So we hope said decisions will generally be reasonable.

[Honors Spectral Image She/her]

I like this rule.

[bedtime]

How about an AI only thread/forum for all AI content?

[Chaos he/him]

4 hours ago, bedtime said:

How about an AI only thread/forum for all AI content?

For what purpose or goal?

Seems like that would just lead to a ton of spam, which also undesirable. 

[bedtime]

8 hours ago, Chaos said:

For what purpose or goal?

Seems like that would just lead to a ton of spam, which also undesirable. 

The goal at the time was simply curiosity which started in this thread: 

The OP of the thread above made a trailer for WoK, and I commented that it would be interesting to see what AI would create if fed that script. The OP seemed curious too.

Along with curiosity, I think it would be fun to talk about the results and maybe even have polling threads for best AI produced trailers/scenes? There are already AI videos that are very decently done, so we know good AI output is possible. And if it's not, we get to laugh at the attempt.

Keep in mind that it's likely that real artists are already spamming their human drawn work on this forum as well as in Sanderson's YouTube videos. In one of Sanderson's videos, Sanderson chats with an artist who drew a picture (not sure if it was Shallan or some other lady as I saw the video weeks ago). Brandon comments on the fact that both her hands were fully exposed—and, storms, without so much as even a glove! Anybody who has read SA knows that she covered her safehand in her sleeve or at the very least wore a glove. The artist didn't even seem embarrassed or concerned about this mistake, and chances are that this artist didn't even read the Stormlight Archives; he was just trying to get his name out and likely does this for other books he knows little to nothing about. He likely doesn't even participate in chatting with any of the forum members or community. That's spam.

You might be questioning if allowing AI content would result in an influx of trolls onto the forum. I can't say for sure, but I highly doubt it. I'm thinking this forum is too niche to draw a net big enough to make much of a difference in spam/trolls by simple allowing AI. I'd be more worried about people using this forum as a platform to sell their own work (such as real artists)—who don't even read the actual novels—than people who are already interested in the books, genuinely love them, and who happen to be able to create AI works.

All I'm talking about is adding one forum/thread. If somehow got out of control, then it could be easily revoked or archived. If you really wanted to make sure trolls didn't just join the forum to post AI content, you could invoke a minimum post count to post AI (perhaps 100-200 posts?). You'd get real book lovers posting it this way.

Btw, thanks so much for answering me and allowing conversation on the topic.

[Chaos he/him]

It's not that the AI content is "trolling", it's that it's trivial to generate. Yes, this is a niche community, but as you can see from some forums, some people can make a lot of posts. Imagine how much can be generated with basically no effort. Sure, I feel like a post like the one you suggest, it's kind of whatever. Maybe a worth a laugh. If even 2-3 people are doing stuff like that often, will people to care to read? I'm not sure. That's the spamminess I'm talking about. The easier it is to do, the more it's like, "Do I want to read or watch something AI generated among all of this AI stuff when I could do something else?" It's a forum that if it exists I think quickly devolves into something not useful.

But, of course, I'm sure there's different opinions on that, and fair enough. I think we as staff are more concerned with the fact that these LLMs are effectively stealing copyrighted information to generate this stuff, which feels... less great. Among other reasons. So, that is a motivation of the AI policy. We don't feel the benefit outweighs that. 

6 hours ago, bedtime said:

Keep in mind that it's likely that real artists are already spamming their human drawn work on this forum as well as in Sanderson's YouTube videos. In one of Sanderson's videos, Sanderson chats with an artist who drew a picture (not sure if it was Shallan or some other lady as I saw the video weeks ago). Brandon comments on the fact that both her hands were fully exposed—and, storms, without so much as even a glove! Anybody who has read SA knows that she covered her safehand in her sleeve or at the very least wore a glove. The artist didn't even seem embarrassed or concerned about this mistake, and chances are that this artist didn't even read the Stormlight Archives; he was just trying to get his name out and likely does this for other books he knows little to nothing about. He likely doesn't even participate in chatting with any of the forum members or community. That's spam.

Again, I think we might have different definitions of "spam". I think there's a pretty colossal difference between someone creating hundreds of images in moments and posting them all vs. an artist who is creating things themselves. There's a pretty important bottleneck there, in that creating art takes time, that inherently makes it far less spammy. 

I can't comment on the video you're talking about, as I do not know the video or the specific art piece therein. We do feel it is very important that 17th Shard is a place that is a pleasant and positive place to be for artists. This is why we have a No Unsolicited Critique policy. We want to be encouraging to artists. Some are doing things for funsies. Some want to improve, some aspire to be professionals. Canonicity is not required nor would be a thing we wish to judge or adjudicate. Not all are looking for feedback like, "Cover the safehand." That might not be the focus of the art piece or the artistic intent. That's not our domain to judge, and if people are not requesting critique, then that is not permitted to judge here, either. 

I think your concern is quite overblown. There's, what, 2000 images under Stormlight Archive Art across nearly ten years? I'm... not really concerned with that pace, I have to be honest. I am also not concerned with some... nefarious artist who is drawing noncanonical Stormlight art in an effort to get their name out, and is using this site to do that. Let's be perfectly real: there is vastly more cosmere art on Instagram. They'd use that. If it wasn't reasonably good, then it wouldn't get traction anyway. And, is there some more nefarious intent? I feel like that's literally never happened on this site. Again, I can't speak to the video you saw, but I would need more information to make such judgments.

Lastly, artists posting their art here without discussing much here is fine. I like seeing cool art here. We are also fine with cosmere content creators posting their videos here. This doesn't happen too often--again, the time required to create such content is an important limiter--and certainly that could be spammy, if a creator was posting their video in a ton of forums. But one topic every once and a while? Sure. If we disallowed that and required all posters of such content to be active on the forums, we would be even less of a Brandon Sanderson hub than we already are.

The difference with AI is, again, the scale that this can be done compared to the effort. That's my spam concern. It is not my personal main reason the AI policy exists, but it is a reason why I think an AI-only forum would quickly become a ghost town with lots of stuff that probably no one looks at. Your mileage will vary, of course, and that is fine, but I just don't think it's an extremely compelling use case to warrant an exception to the broad policy. 

6 hours ago, bedtime said:

I'd be more worried about people using this forum as a platform to sell their own work (such as real artists)—who don't even read the actual novels—than people who are already interested in the books, genuinely love them, and who happen to be able to create AI works.

This is already highly restricted and moderated here. It sometimes comes up, but has not been a major issue.

[bedtime]

@Chaos

Thanks so much for that reply!

This has put the issue to bed for me in the best way, so please excuse my short reply. Really, all I wanted was to do was see a script turn into an AI video on this forum for the OP, but I hadn't thought of the scope of its implications. Perhaps, having been so focused on my objective, I got a trifle carried away with my points and lacked many considerations...

I really liked your comment on the time vs effort for the artists. This makes sense. Seems that one artist I referred to (the one Sanderson talked to) had a effect on me, but I know that there's a plethora of great Stormlight Archive art.

Anyways, I'm content with this. Loving the job you guys are doing with the forum!