# Updates to the site guidelines, especially regarding gen-AI

**URL:** <https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315>\
**Category:** Meta Discussion\
**Created:** [December 2, 2025, 10:58pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315 "2025-12-02T22:58:56Z")\
**Posts on this page:** 20\
**Page:** 2

<div class="post-metadata">

**Author:** ![kellertuer](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/kellertuer/32/220707_2.png) [@kellertuer](https://discourse.julialang.org/u/kellertuer)\
**Post date:** [September 11, 2026, 12:54pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/21 "2026-09-11T12:54:41Z")

</div>

You have to take my word for exactly nothing. I would consider myself one of the most unimportant people concerning running the forum.

I am tired of this. I tried to help, I think I achieved the opposite, so I leave it be.  
But I can guarantee that posts that I write here have near-zero polishing and absolute zero AI involvement. They stem from pure physical keyboard actions letter by letter.

---

<div class="post-metadata">

**Author:** ![gwr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/gwr/32/33632_2.png) [@gwr](https://discourse.julialang.org/u/gwr)\
**Post date:** [September 11, 2026, 12:57pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/22 "2026-09-11T12:57:55Z")

</div>

> [@kellertuer](#):
>
> They stem from pure physical keyboard actions letter by letter.

That is an interesting proof of quality. You may take my honest word for it that the last post, while lengthy, has been written and typed by me entirely. If it reads polished than maybe because I used to have 119/120 TOEFL. 😜

---

<div class="post-metadata">

**Author:** ![kellertuer](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/kellertuer/32/220707_2.png) [@kellertuer](https://discourse.julialang.org/u/kellertuer)\
**Post date:** [September 11, 2026, 12:59pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/23 "2026-09-11T12:59:47Z")

</div>

Uh, I never did a TOEFL – but that is also off topic.

---

<div class="post-metadata">

**Author:** ![goerz](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/goerz/32/3269_2.png) [@goerz](https://discourse.julialang.org/u/goerz)\
**Post date:** [September 11, 2026, 1:04pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/24 "2026-09-11T13:04:09Z")

</div>

> [@gwr](#):
>
> Currently, there is a _strict_ rule (e.g., only minor editing and pure translation)

You completely made up this rule that you’re complaining about. The moderators have told you that this rule does not exist. Now you’re still complaining that you don’t get a vote about this non-existent rule

> [@gwr](#):
>
> the moderators have a lot of discretion to decide what is acceptable and what is not.

Well, yeah, that’s what makes them moderators. _Somebody_ has to decide at the end of the day whether some particular post crosses the line to LLM slop or not. They are reasonable people. You can talk to them.

> [@gwr](#):
>
> You cannot participate in a forum called “Discourse” not because you provided unhelpful or false arguments in a discussion, but because you used a prohibited tool and your text was too polished and a tad too long.

That’s the exact opposite of what the moderators have told you. There are no “prohibited tools”. Post do not get censored because they are “too polished or a tad too long”.

---

<div class="post-metadata">

**Author:** ![mbauman](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mbauman/32/31082_2.png) [@mbauman](https://discourse.julialang.org/u/mbauman)\
**Post date:** [September 11, 2026, 1:08pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/25 "2026-09-11T13:08:51Z")

</div>

> [@goerz](#):
>
> They are reasonable people. You can talk to them.

There’s a second half to discourse, too, and that’s _listening_. Please take care to read what I’ve been saying, too.

---

<div class="post-metadata">

**Author:** ![mbauman](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mbauman/32/31082_2.png) [@mbauman](https://discourse.julialang.org/u/mbauman)\
**Post date:** [September 11, 2026, 1:13pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/26 "2026-09-11T13:13:22Z")

</div>

I do think we can improve some of the site communication around this. @Mason made a good suggestion last year that we never took action on, and @dilumaluthge discovered that we can detangle the **Spam or Generative AI** flag into two! When I made the change, I don’t think the ability to add an additional flag had percolated down to our instance yet (or maybe I had just missed it).

But having better messaging on the flags and resulting canned moderation direct messages should help, I would think.

---

<div class="post-metadata">

**Author:** ![gwr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/gwr/32/33632_2.png) [@gwr](https://discourse.julialang.org/u/gwr)\
**Post date:** [September 11, 2026, 1:13pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/27 "2026-09-11T13:13:34Z")

</div>

> [@goerz](#):
>
> You completely made up this rule that you’re complaining about. The moderators have told you that this rule does not exist.

The guidelines say:

> - Don’t post generative AI output (but _direct human language translation_ and minor editing is ok).

I am not making anything up. If you want to imply that being _output_ is what counts, then I find that sophistic.

> [@goerz](#):
>
> They are reasonable people. You can talk to them.

Yes, I know that and I honestly appreciate that, which may have been lost in my persistent argumentation, but that is truly the case. Nevertheless, as written above, I very much prefer very clear rules that allow to use generated output if it is endorsed and has reasonable quality.

> [@goerz](#):
>
> There are no “prohibited tools”. Post do not get censored because they are “too polished or a tad too long”.

Yes there are, read the rules. My post has been “banned” because I disclosed the use of AI (thus next time I should probably be more clever and dishonest), but nobody could really tell me where it was “slop” or whether it was not “my voice” or “my message” and whether the content was not helpful and clearly related to the discussion?

---

<div class="post-metadata">

**Author:** ![gwr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/gwr/32/33632_2.png) [@gwr](https://discourse.julialang.org/u/gwr)\
**Post date:** [September 11, 2026, 1:15pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/28 "2026-09-11T13:15:34Z")

</div>

> [@mbauman](#):
>
> There’s a second half to discourse, too, and that’s _listening_. Please take care to read what I’ve been saying, too.

I do and please return the favor as well.

EDIT: This is exactly, why GenAI can be helpful, because it greatly helps to avoid misunderstanding in communication as the bandwidth of written communication is unfortunately quite narrow.

---

<div class="post-metadata">

**Author:** ![gwr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/gwr/32/33632_2.png) [@gwr](https://discourse.julialang.org/u/gwr)\
**Post date:** [September 11, 2026, 1:23pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/29 "2026-09-11T13:23:04Z")

</div>

> [@mbauman](#):
>
> But having better messaging on the flags and resulting canned moderation direct messages should help, I would think.

That might indeed go into the right direction. I would simply suggest to a) appeal to refrain from AI use in direct communication with people (e.g., more or less “chatting” or “talking” with each other), b) make clearer content quality related guidelines not simply (do not use a certain tool), and c) clearly offer the poster a chance to _heal_ the post.

What about that?

EDIT: A more technical solution would be “transcribed video messages” so there can be no doubt; might even be a market for such a tool. 😉

---

<div class="post-metadata">

**Author:** ![mbauman](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mbauman/32/31082_2.png) [@mbauman](https://discourse.julialang.org/u/mbauman)\
**Post date:** [September 11, 2026, 1:33pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/30 "2026-09-11T13:33:23Z")

</div>

Holy cow, I think we might have done it! @gwr: that’s the status quo, or at least the intent of it! 🙂

If and how we improve the wording in the guidelines, flags and DMs is really the only question… along with the fact that I’d say that this _entire site_ is for direct communication with people.

But, yes, that is exactly our intention.

---

<div class="post-metadata">

**Author:** ![goerz](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/goerz/32/3269_2.png) [@goerz](https://discourse.julialang.org/u/goerz)\
**Post date:** [September 11, 2026, 1:35pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/31 "2026-09-11T13:35:47Z")

</div>

> [@gwr](#):
>
> The guidelines say:
> 
> > - Don’t post generative AI output (but _direct human language translation_ and minor editing is ok).
> 
> I am not making anything up.

But the moderators have told you that you are misreading this rule! This rule was formulated in the pre-agentic ara, when we still had ChatBots. It doesn’t say “don’t post anything that has ever been touched by an LLM”. It doesn’t say “_only_ direct language translation and minor editing”. That is an exemplary, not a complete list. And I think “minor editing” can do a lot of work here. I’m all for rewriting that rule to

> Don’t post unfiltered generative AI output (but, e.g., direct human language translation and refinement is ok).

if that alleviates your concerns.

> [@gwr](#):
>
> If you want to imply that being _output_ is what counts, then I find that sophistic.

Of course it’s the output that counts! Why do think the rule exists in the first place? It’s not because “LLM bad!”. It’s because LLMs have the ability to generate vast quantities of low quality text that can easily overwhelm the discussion. That “low quality” is the entire issue, and the rule exists to try to mitigate that.

---

<div class="post-metadata">

**Author:** ![gwr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/gwr/32/33632_2.png) [@gwr](https://discourse.julialang.org/u/gwr)\
**Post date:** [September 11, 2026, 1:37pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/32 "2026-09-11T13:37:28Z")

</div>

> [@mbauman](#):
>
> Holy cow, I think we might have done it!

… only reasonable people use Julia. 😉

> [@mbauman](#):
>
> along with the fact that I’d say that this _entire site_ is for direct communication for people.

I beg to disagree a bit, because given that we by now produce Julia software and even Julia language itself using GenAI, it seems a bit too harsh to prevent GenAI in producing posts _at least partially_.

GenAI support in my opinion must not necessarily lead to unpersonal or indirect communication, if it is carefully crafted under supervision and authorized. But hey … 🙂

---

<div class="post-metadata">

**Author:** ![goerz](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/goerz/32/3269_2.png) [@goerz](https://discourse.julialang.org/u/goerz)\
**Post date:** [September 11, 2026, 1:38pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/33 "2026-09-11T13:38:12Z")

</div>

> [@gwr](#):
>
> GenAI support in my opinion must not necessarily lead to unpersonal or indirect communication, if it is carefully crafted under supervision and authorized.

We all agree on that!!!

---

<div class="post-metadata">

**Author:** ![gwr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/gwr/32/33632_2.png) [@gwr](https://discourse.julialang.org/u/gwr)\
**Post date:** [September 11, 2026, 1:39pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/34 "2026-09-11T13:39:31Z")

</div>

> [@goerz](#):
>
> That “low quality” is the entire issue, and the rule exists to try to mitigate that.

Then please clearly show me where my banned post showed low quality and was a mere data dump! That’s the entire issue for me. 😉

EDIT: That may even be used for _case-based learning_, e.g., a collection of banned posts with clear reasoning: This was banned because XYZ. Where XYZ are clear quality markers, not just “that’s an English word last used by Lincoln.”

---

<div class="post-metadata">

**Author:** ![goerz](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/goerz/32/3269_2.png) [@goerz](https://discourse.julialang.org/u/goerz)\
**Post date:** [September 11, 2026, 1:49pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/35 "2026-09-11T13:49:29Z")

</div>

Since I’m not a moderator, I don’ know which post of yours was banned, or whether it would have crossed any of _my_ lines. But if your complaint is about some particular post of yours that you think was flagged improperly, that’s probably not something that impacts this _general_ discussion of the rules. You should engage on that one-on-one with the moderator that banned the post in a direct message.

**Edit:**

> [@gwr](#):
>
> My post has been “banned” because I disclosed the use of AI (thus next time I should probably be more clever and dishonest)

Well, yeah, okay: Don’t disclose the use of AI! You don’t have to, and the intent of the rules is such that if you’re using AI responsibly, there shouldn’t be any need to disclose it, because you’re simply making well-written posts that _you_ stand behind. I don’t think it’s “dishonest”. Maybe your post would have been totally fine. But if you preface it with “This post was written by AI” when there’s a guideline that says “Don’t post AI”, then _of course_ it’s going to get flagged.

---

<div class="post-metadata">

**Author:** ![gwr](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/gwr/32/33632_2.png) [@gwr](https://discourse.julialang.org/u/gwr)\
**Post date:** [September 11, 2026, 1:52pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/36 "2026-09-11T13:52:23Z")

</div>

Yes, I did and was not given a “low quality” indication clearly, e.g., completely unrelated, impersonal, not a human voice etc.

You claimed that there is no problem and that “low quality” were the only issue, I just cited a counterexample here to refute that claim.

---

<div class="post-metadata">

**Author:** ![kellertuer](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/kellertuer/32/220707_2.png) [@kellertuer](https://discourse.julialang.org/u/kellertuer)\
**Post date:** [September 11, 2026, 1:56pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/37 "2026-09-11T13:56:11Z")

</div>

> [@goerz](#):
>
> > [@gwr](#):
> >
> > GenAI support in my opinion must not necessarily lead to unpersonal or indirect communication, if it is carefully crafted under supervision and authorized.
> 
> We all agree on that!!!

Yes we indeed do.  
I am so annoyed by AI interaction that I will not use it in communication (just a bit of bug fixing and generating some TeX code) – but on that other people of course might have other opinions.

---

<div class="post-metadata">

**Author:** ![mbauman](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/mbauman/32/31082_2.png) [@mbauman](https://discourse.julialang.org/u/mbauman)\
**Post date:** [September 11, 2026, 2:07pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/38 "2026-09-11T14:07:30Z")

</div>

The Discourse moderation tools are wildly opaque — to admins, mods, and users alike. And the direct messages the system sends were written a decade ago. I actually don’t think they’ve aged well. A message from @system that says “This is an automated message from Julia Programming Language to let you know that your post was hidden” sure sounds more agentic today than it did ten years ago.

But yes, all such direct messages include a paragraph like:

> This post was hidden due to flags from the community, so please consider how you might revise your post to reflect their feedback. **You can edit your post after 10 minutes, and it will be automatically unhidden.**

@gwr I just checked and I see that the message you got sent in that particular case was different. I don’t understand why. It shouldn’t be different. It’s quite frustrating to me and I can understand how it’d frustrate you too.

It’s also very unfortunate that responding to or even flagging those @system messages are completely unmonitored and invisible (or barely visible, not triggering the typical notifications) to moderators.

Anyhow, we can improve most of these things.

---

<div class="post-metadata">

**Author:** ![Benny](https://avatars.discourse-cdn.com/v4/letter/b/49beb7/32.png) [@Benny](https://discourse.julialang.org/u/Benny)\
**Post date:** [September 11, 2026, 4:19pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/40 "2026-09-11T16:19:14Z")

</div>

> [@gwr](#):
>
> You guys fix this like good AIs

I get you don’t think you’re getting your way, but why dehumanize people actively trying to improve your experience?

> [@mbauman](#):
>
> please consider how you might revise your post to reflect their feedback.

While we’re criticizing the moderation system, are there good ideas for adding more information to feedback? I’m guessing encouraging arbitrary negative responses are probably a bad idea, but _this_ is what I got for the earlier parody.

> **Expand for exerpt.**
>
> Your post was flagged as **off-topic or unfocused**. The community feels it is not a good fit for the topic, as currently defined by the title and the first post, or that the topic itself lacks enough focus to facilitate a productive conversation.
> 
> This post was hidden due to flags from the community, so please consider how you might revise your post to reflect their feedback. **You can edit your post after 10 minutes, and it will be automatically unhidden.**
> 
> However, if the post is hidden by the community a second time, it will remain hidden until handled by staff.

Basically just a generic “off-topic”. I’m betting the people who flagged it had a lot more to say than that. Maybe it’s not time for jokes, it’s too rude, etc, and probably not a suggestion to make the joke more relevant somehow. They could DM me, but ideally there’s a convenient middle ground (with an opt-out) between generic flags and several full conversations.

---

<div class="post-metadata">

**Author:** ![Benny](https://avatars.discourse-cdn.com/v4/letter/b/49beb7/32.png) [@Benny](https://discourse.julialang.org/u/Benny)\
**Post date:** [September 11, 2026, 6:55pm UTC](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315/42 "2026-09-11T18:55:43Z")

</div>

Wouldn’t be the first joke to not land as intended in this thread, as the rest of my previous comment described. If I had less patience, I may have decided to flag that as well, so more feedback options with flags would be nice.

[Previous page](https://discourse.julialang.org/t/updates-to-the-site-guidelines-especially-regarding-gen-ai/134315.md?page=1)