PathfinderWiki talk:Policies and guidelines

Archives:

1 (2009-2014)

Remove defunct policies from this page (RESOLVED)

This page's "List of policies" section should not display the outdated PathfinderWiki:Pathfinder Society scenarios, which exists only for archival purpose. — Descriptivist (talk) 09:47, 7 August 2024 (UTC) -Oznogon (talk) 16:04, 8 August 2024 (UTC)

Community or content policy

There are two separate recent discussions that might warrant a content or community policy to define acceptible uses of content.

• On the wiki Discord, Rudy asked on 29 March 2025 (today):

“

Are AI tools like chatGPT being used to rephrase copy-pasted paragraphs and post them on PathfinderWiki, and if not, why?

”

The ensuing discussion indicated a strong consensus among editors and administrators that the use of large language models (LLM) or other forms of generative artificial intelligence (AI) are unwelcome on the wikis. This is in line with Paizo's own stance against the use of generative AI in their products, or in works published to venues they control, like Pathfinder and Starfinder Infinite.

The Community Use Policy (CUP) itself does not codify a specific prohibition against the use of LLMs or AI.

• The discussion (coincidentally also opened today, but in reference to actions in late 2024) at Project talk:Images regarding the User:Descriptivist's removal of artwork deemed to be not safe for work (NSFW).

The CUP provides a relevant provision (underlined emphasis mine):

“

Don't do anything that might hurt or damage Paizo. You agree to use your best efforts to preserve the high standard of our intellectual property. You agree to not use this permission for material that the general public would classify as "adult content," offensive, or inappropriate for minors, and you agree that such use would irreparably harm Paizo. You agree to not use Paizo's trade dress—that is, you may not make your material look like ours. You agree that such use would irreparably harm Paizo.

”

However, the artwork in question is itself released to our use under the CUP, so it's unclear whether this provision is sufficiently strict for the wiki's audience.

These concerns suggest a potential need for a content policy or guideline, similar in theme if not content to the Fandom community guidelines, that specifies restrictions and preferences for the nature of PathfinderWiki's content.

I'm not privy to or aware of any legal concerns about content moderation that the recent Tabletop RPG Historical Society nonprofit now charged with managing the wiki faces. If any exist, a content policy could also provide a venue to define such legally required notices, definitions, or accommodations. -Oznogon (talk) 01:28, 30 March 2025 (UTC)

Repeating what I commented on the NSFW template talk page.

I strongly oppose the removal of "NSFW" images, especially considering none of the artwork depicts genitals or full nudity. It is official Paizo artwork, has been present on the website for over a decade, shows off no more than you'd see on a beach, many creatures have few if any alternative images available, and the definition of NSFW is highly subjective.

Does File:Basileus.jpg also count as NSFW? How about File:Ankana.jpg and File:Seoni tattoo.jpg? I can easily see someone considering these examples and many more as NSFW. Regardless, intentionally hiding or obscuring some part of the world of Golarion goes directly against the purpose of the PathfinderWiki.

I would fully agree with the CUP provision regarding this matter if the artwork came from a third party, rather than Paizo themselves. --Rexert (talk) 12:29, 30 March 2025 (UTC)

Generative AI Policy

Starting a subheader section here for discussion of this point specifically so we can keep the two discussions distinct within the overall content policy discussion.

While I agree with what some said on the Discord that we may never know if someone is using generative AI to create wiki articles, I think the wiki benefits from letting users (both editors and readers) know that a human has created all content and vetted it for accuracy. This encourages editors to edit, because they know their contributions are valued and not going to be superseded by some robot rehash of canon sources, and adds a level of trust for readers who know that some algorithm hasn't put fake info up here because it couldn't find the actual answer.

Additionally, Paizo has come our very publicly against generative AI, and I think setting a policy that doesn't do the same is a betrayal of a tacit bond between the company and the community. As a member of the wiki staff, I want to make sure Paizo knows that we respect their wishes not to willfully put their copyrighted material into an LLM, even if we feel it's inevitable that others will. As a member of Paizo's staff, and one who has to send DMCA takedowns as a part of my everyday job, I can't approve any policy that would encourage or even allow such use of the company's IP.

Also, I have asked a number of LLMs all sorts of questions about Pathfinder, Starfinder, and Paizo just to see what it could get right, and I'd say it was batting sub-.500 on even basic questions that you could find through a non-LLM google search. It's simply not something we should introduce into a resource that is supposed to value accuracy.— Yoda8myhead (talk) 21:39, 31 March 2025 (UTC)

Agree. It feels like there would be (or already is, in off-wiki venues) a strong consensus among editors against the use of LLMs/AI to edit, paraphrase, or generate content, and it does sound like we need to codify it as a restriction on the wikis. -Oznogon (talk) 22:34, 31 March 2025 (UTC)