Consider how to handle AI generated comments and posts

[yorkshire-pudding]

In the forum we have started to see some comment posts that have several indicators of being wholly written by AI. While we generally think it should be acceptable for someone to include a suggestion, script or solution from AI, it should be attributed in the same way that if you got a solution from the Drupal issue queue or Stack Exchange or a web development resource, then we would expect that person to credit their source and not pretend it is their own work. Also, a comment should not be wholly AI but should include explanation. We can't verify if any or all of these are correct without substantial effort and there is a risk that the forum becomes less useful over time.

We discussed in the Dev Meeting on 2025-10-23 https://www.youtube.com/live/fbaJlhcxL1g?si=xgDhhWirvKdX0AMm&t=351 and a number of ideas were floated including:

• A checkbox for comments that will mark a comment as AI. There isn't yet a standard for marking up content as AI generated but there are some proposals:
  • https://aimetatag.org/ (this was found by @quicksketch and discussed in the meeting)
  • Proposal: Meta Tag for AI Generated Content whatwg/html#9479 (a suggestion in the queue for the HTML standard)
• A reputation system so new accounts could be flagged/limited in some way. @cellear has built something like this in the past and may be able to share how such a mechanism could work.
• @keiserjb suggested that AI could help us detect and help manage this. It would be good to understand how that would work in the forum.

[NormPlum]

I suggest there be a way for authors to mark their post as AI-generated/enhanced, anyone to flag a post as suspected AI-generated/enhanced, and then a way to visually highlight marked/flagged posts.

[indigoxela]

Progress in a chat thread: we might be able to come up with a way simpler solution, that doesn't require any additional fields or workflows or modules. 😉 See discussion details: https://backdrop.zulipchat.com/#narrow/channel/218769-Forum/topic/Add.20some.20sort.20of.20AI.20guideline

[olafgrabienski]

I suggest there be a way for authors to mark their post as AI-generated/enhanced, anyone to flag a post as suspected AI-generated/enhanced, and then a way to visually highlight marked/flagged posts.

I like the combination of a checkbox and a flag. How do we get an idea about the question of how many flags a post needs to have before it is marked as AI generated?

And maybe we find a way to not highlight such posts visually, by marking and de-emphasizing them somehow at the same time.

[olafgrabienski]

Progress in a chat thread: we might be able to come up with a way simpler solution, that doesn't require any additional fields or workflows or modules.

Just for reference, one of the ideas discussed in Zulip, to add some text to the Forum's About page, has already been implemented: https://forum.backdropcms.org/about