# Expressly forbid 'AI' generated content **URL:** https://discuss.privacyguides.net/t/expressly-forbid-ai-generated-content/26420 **Category:** Meta **Tags:** discourse **Created:** 2025-04-03T18:06:31Z **Posts:** 39 ## Post 1 by @anon42475305 — 2025-04-03T18:06:31Z The Privacy Guides forum should adopt a blanket policy expressly forbidding the use of generative ‘AI’ except for the purposes of machine translation. Justification: - ‘AI’ generated content arguably already falls foul of many existing rules and their spirit. For example: “Post Only Your Own Stuff”, “Improve the Discussion”, etc. - ‘AI’ generated content often contains factual inaccuracies which introduced an unnecessary additional burden for moderators - ‘AI’ generated ‘evidence’ is often used as a substitute to doing sound research and providing verifiable sources * * * Basically I’ve recently noticed an uptick in the amount of ‘AI’ generated content and I cannot recall a single time that content has genuinely been useful. In almost every such situation a human written response would not only have been more helpful, but also would’ve contributed far more positively to the forum as a whole. --- ## Post 2 by @KathyM — 2025-04-03T19:02:23Z AI also creates poisoned data though. Kinda useful for fighting the AI surveillance software. --- ## Post 3 by @anonymous309 — 2025-04-03T19:04:32Z I think, in this case, that would be counter to the point of forum members trying to help inform people about privacy and security. I know for me, I don’t need anymore help getting things wrong :joy: --- ## Post 4 by @anon36940904 — 2025-04-03T19:28:15Z This one will be tricky. I do believe AI generated comments and info must not not exist on this forum and less than educated folks can seem to be more than what they really are when it comes to topics of this forum and PG. How to go about it is the next big question. Luckily for us, as far as I know, only @jonah and team are responsible for it so hope they deliberate and come up with a system that mitigates this issue. --- ## Post 5 by @anonymous315 — 2025-04-03T19:53:13Z Having recently flagged a handful of these, it’d be nice if the iron fist slammed down and banned AI generated content. Other forums have already taken this stance. --- ## Post 6 by @xe3 — 2025-04-03T19:55:21Z > **Lighthearted over-caffeinated satire** > > :glowing_star: **Re: Proposal for AI Content Policy** :glowing_star: > > Hey everyone! :blush: > > I totally see where you’re coming from with your concerns about generative AI content in our forum. Here are my thoughts on the matter! :thought_balloon: > > - **Quality Over Quantity** :chart_increasing: > > - **Factual Accuracy** :white_check_mark: > > - **Encouraging Research** :books: > > - **Community Engagement** :handshake: > > While I see the potential for AI in certain contexts, like machine translation, I think your proposal could help maintain the quality and spirit of our forum. Let’s keep the conversation going! :rocket::sparkles: > > Cheers! :tada: * * * > **My actual (initial & conflicted) thoughts** > > I’d have to think harder about my feelings towards a blanket ban. But, like you, I’ve been feeling that some people’s overreliance on AI has had a small but noticeable negative impact on the forum. In particular I’ve noticed two negative trends: > > 1. _“I asked “AI” and it told me: \_\_\_\_”_ (as a response to a question, or as an assertion/statement of “fact”) > > 2. Obvious (and not so obvious) copy pastes of AI generated text or “slop” (like my satirical example) without any disclaimer and without putting it in quotes and attributing it to AI. > > With that said, I’m not sure that outright banning AI in full is necessarily the right approach (I’m also not sure that it isn’t). > > I think straight up copy pasting AI generated outputs as if they are your own should be banned. But I can also envision some less clear cut edge cases where AI use in limited ways, isn’t necessarily a bad thing or at least doesn’t rise to the level of badness requiring an outright ban. (e.g. using AI to correct tone or grammar, or to format posts for someone who struggles to communicate their thoughts clearly/writes unformatted walls of text), or for answering the type of concrete simple technical question that AI tends to be pretty good at (e.g. using it as a supplement to `man` when looking for a specific Linux terminal command). > > I’m kind of just thinking out loud here, I have a somewhat conflicted point of view with respect to LLMs. LLMs are useful and convenient for various things, but the proliferation of AI generated content is contributing to making the internet a worse place in a lot of ways, and may further threaten our ability to focus, be thoughtful and deliberative, do research, and to verify/check sources (all of which are already growing problems in my eyes, even before AI came around). > > At a minimum I think any AI generated text in comments should have a clear disclaimer, be formatted as a quotation, and be kept to a minimum. And I wouldn’t necessarily be opposed to an outright ban, that might be the best approach, I’m just unsure about it. --- ## Post 7 by @anon36940904 — 2025-04-03T19:56:09Z Damn. I don’t know how you are always more thoughtful than everyone. I envy you. --- ## Post 8 by @KevPham — 2025-04-03T19:59:08Z I was under the presumption that AI or LLM-generated content was already banned here. We can do a better job of communicating that though in our rules. That follows whether or not our existing rules deserves an overhaul. I am happy to hear any suggestions on specific wording though --- ## Post 9 by @anon42475305 — 2025-04-03T19:59:53Z > [@xe3](#): > > using AI to correct tone or grammar, or to format posts for someone who struggles to communicate their thoughts clearly/writes unformatted walls of text I agree with what your saying, but I disagree that it requires an exception. The reality is that if I can’t tell you’re using ‘AI’ to fix your grammar then it will never become an issue. Carving out an exception for that is only confusing. The issue I have is with content that is obviously ‘AI’ generated and I do think AI formatted content should be included in that. Like with your satirical example, even if my text remains 99% the same formatting can also be extremely abnoxious. See below: > **Example** > > I agree with what you’re saying :smiley:, but I must respectfully disagree that it requires an exception :thinking:. The reality is that if I can’t tell you’re using ‘AI’ to fix your grammar :writing_hand:, then it will never become an issue :sweat_smile:. Carving out an exception for that is only confusing :face_with_spiral_eyes:—you know? > > The issue I have is with content that is obviously ‘AI’-generated :man_shrugging:, and I do think AI-formatted content should be included in that :grimacing:. --- ## Post 10 by @anon63378630 — 2025-04-03T20:10:13Z > overreliance I recently attended a talk for beginners and the speaker kept repeating along the lines “if you need help/elaboration on this just ask ChatGPT”. It saddens me to see this become the norm. AI tools can be useful but they should not be constantly relied upon. Using them to generate forum responses is utterly useless and brings nothing to the conversation but verbose statements and potential misinformation. It also doesn’t necessarily help you truly understand the issue/problem at hand. --- ## Post 11 by @anon42475305 — 2025-04-03T20:11:47Z > [@KevPham](#): > > I was under the presumption that AI or LLM-generated content was already banned here. We can do a better job of communicating that though in our rules. I kind of was too but I realised it has become a recurring issue recently and I’ve had to flag / remove many such posts. I personally appreciate the rules being vague -ish guidelines rather than carved in stone but clearly prohibiting all use of generative ‘AI’ is necessary imo. --- ## Post 12 by @xe3 — 2025-04-03T20:32:26Z > [@anon42475305](#): > > I agree with what your saying, but I disagree that it requires an exception You are probably right. (And my recollection is that @KevPham is correct, I think there is technically already an unofficial or official rule against AI generated content in posts/comments. At least I recall @jonah saying something along those lines). (edit: [this is the exchange I was remembering](https://discuss.privacyguides.net/t/ai-generated-text-policy/18489/3)) > The reality is that if I can’t tell you’re using ‘AI’ to fix your grammar then it will never become an issue. Just to be clear, I don’t personally do this (not for tone, not for grammar, not for anything) My comments are my own (as evidenced by my often embarrassing grammar, meandering comments, and the occasional made up word, + the 13 edits it sometimes takes me to try to clarify what I meant to say) :grinning_face: Like you, I find the generic wishy-washy AI tone, overuse of emojis, relentless bland cheerfulness to be annoying, distracting and not helpful. > I do think AI formatted content should be included in that While I’m often annoyed by AI generated text, I think _decent_ AI generated text can in some cases clarify _bad_ human formed text. We’ve all encountered people online who don’t use paragraphs, don’t use punctuation, are really passionate about a subject but struggle to communicate effectively or calmly, or write in an almost stream-of-consciousness style. That is the sort of edge case I was thinking of. But you are probably right that making explicit exceptions for random edge cases is an unnecessary over-complication. --- ## Post 13 by @mountainofmeat — 2025-04-03T20:41:50Z If I wanted to know what AI has to say I’d just ask AI myself. It sometimes feels similar to people “shilling” for a product they really like so they keep bringing it up in topics remotely related to it. So in that sense a full ban seems too strict, we (luckily) haven’t reached the point yet where one AI user is talking to another AI user. Telling people why they shouldn’t use AI would hopefully be more useful in the long run, but maybe that’s too optimistic. Edit: There also won’t be a good way to combat it if it improves to a point where you can’t tell it apart anymore, so people should learn to use it like any other tool instead of overrelying on it and feeling like they are part of a revolution of never-thinkers. --- ## Post 14 by @anon42475305 — 2025-04-03T20:52:36Z > [@xe3](#): > > While I’m often annoyed by AI generated text, I think _decent_ AI generated text can in some cases clarify _bad_ human formed text. > > We’ve all encountered people online who don’t use paragraphs, don’t use punctuation, are really passionate about a subject but struggle to communicate effectively or calmly, or write in an almost stream-of-consciousness style. That is the sort of edge case I was thinking of. But you are probably right that making explicit exceptions for random edge cases is an unnecessary over-complication. Yeah, I get where you’re coming from, but again, that’s a situation where you could probably get away with discreetly using ‘AI’ to correct your style without ever running into trouble with it. Honestly, though, I would strongly advise such people to take the effort to learn punctuation, paragraphs and general formatting. ‘AI’ can be used to help you learn these concepts, but research already shows how detrimental reliance on ‘AI’ can be for people’s ability to complete simple tasks since they constantly outsource that minor effort. I’m far from perfect at punctuation, especially commas lol, but I really don’t think it requires an unreasonable amount of effort to format text somewhat coherently. --- ## Post 15 by @xe3 — 2025-04-03T21:10:31Z I think I’m no longer on the fence. I do support: > Expressly forbid _‘AI’ generated content_ * * * > **bikeshed** > > > Honestly though, I would strongly advise such people to take the effort to learn punctuation, paragraphs and general formatting > > I think that is good advice most of the time, but in many cases the issue isn’t _learning these concepts,_ in many cases it has more to do with mental differences (which doesn’t mean they shouldn’t still focus on building good clear communication skills, and isn’t an excuse in cases of apathy/laziness, but in some cases they also face greater hurdles than most people wrt expressing themselves effectively). --- ## Post 16 by @mountainofmeat — 2025-04-03T21:23:22Z I think pointing out mistakes AI makes can be a good opportunity to teach people why it’s bad. If you just hate the writing style but every “fact” it states is perfect, it’s really not different from someone rambling on for multiple pages but not making any specific errors in their reasoning. People are lazy (including me and you if we’re being honest), they trust their favorite news host, influencer, random people on social media and finally AI, all it takes is that it feels like a figure of authority on the subject. A ban will just reinforce their feeling that they are being excluded and they’ll keep going into the cult of stupidity that tells them it’s not cool to be skilled or talented or informed. > **Sidenote** > > The reason I say it like that is because common sentiment with AI “artists” (which I hate like the scum of the earth) seems to be that spending time to improve ones skill is a waste of time when you can just write a prompt and call yourself an artist. Most people don’t care because they honestly cannot tell, they just think it’s kinda mediocre art. All of that said, I too think consuming AI generated content (so reading text or looking at images) is a complete and utter waste of time, if the poster could not be bothered to spend the time to come up with something of their own, why should I bother to engage with it? --- ## Post 17 by @anon73250778 — 2025-04-03T21:23:41Z I don’t know if the moderators are just so damn good at detecting AI at its posting time but I seem to be bad at detecting AI. We should outright ban it. The internet is bad enough that if we don’t fight against this form of brain rot (letting AI “think” or “elaborate” for you), it will slowly destroy the foundation of the internet and we will have a dead internet because it is run by AI for AI. It is silly to have to use an AI to detect other AI but we are here now. --- ## Post 18 by @jonah — 2025-04-03T23:34:58Z > [@anon73250778](#): > > but I seem to be bad at detecting AI. This generally sums up my problem with creating a rule that forbids AI: > **[Jonah Aragon (@jonah@neat.computer)](https://mastodon.neat.computer/@jonah/114266203829915235)** > > > "What do we do then? Run around accusing people of using AI?" > > @adam@social.lol this is absolutely my biggest concern with such a rule against AI generated content, and it is something that a lot of my colleagues in the education sector are dealing... > [@anon42475305](#): > > ‘AI’ generated content arguably already falls foul of many existing rules and their spirit. For example: “Post Only Your Own Stuff”, “Improve the Discussion”, etc. I would probably argue that most AI content does fall afoul of these rules, yes, and therefore this additional rule you’re suggesting is probably not necessary. I am generally happy with how AI content has been moderated so far, and I don’t really think an additional rule will change anything in how we moderate, nor will it realistically deter the type of people who would post AI content in the first place. Personally, my moderation ethos is basically that too many rules strongly encourages “rules lawyering,” which I don’t really tolerate from people. I prefer to have general expectations of behavior and trust that people can handle themselves, and if they can’t they can find another community :man_shrugging: My final thought: I have never been particularly actively involved in moderation, and I will support the decision that the rest of you come to, but these are my 2 cents on how I like communities to work. --- ## Post 19 by @anon73250778 — 2025-04-04T04:02:16Z At this point I just appreciate what you guys do. Moderating is hard enough that we don’t need AI to vastly complicate things. I can’t imagine how hard it is for the education sector as well. --- ## Post 20 by @asanyan — 2025-04-04T10:29:29Z While I agree that it’s probably futile to ban it entirely due to how hard AI content is to detect, it would be nice to have it as a guideline for people who do care about following the rules… edit: AI is also helpful in obfuscating your writing style against stylometry attacks, so prob don’t ban it --- ## Post 21 by @anon42475305 — 2025-04-04T11:47:45Z > [@jonah](#): > > I would probably argue that most AI content does fall afoul of these rules, yes, and therefore this additional rule you’re suggesting is probably not necessary. That’s fair, as I said, I assumed it was already largely prohibited but I figured if I was unclear on the issue others probably were too. I feel like I’ve received some of the additional clarity I was looking for so I agree that a new rule specifically forbidding ‘AI’ generated content is probably unnecessary. That being said I wouldn’t necessary mind a post similar to the [New posting guideline: Avoid editorialized titles](https://discuss.privacyguides.net/t/new-posting-guideline-avoid-editorialized-titles/25808) post to refer people to / for setting general expectations. --- ## Post 22 by @ignoramous — 2025-04-04T12:24:40Z > [@anon63378630](#): > > kept repeating along the lines “if you need help/elaboration on this just ask ChatGPT For 101, ChatGPT (especially if grounded on facts via attachments or web search), is _very_ handy for CS topics, if nothing else. Or, at least that’s what the students of my course tell me. > [@jonah](#): > > would probably argue that most AI content does fall afoul of these rules, yes, and therefore this additional rule you’re suggesting is probably not necessary Unambiguous in that, it is a singular rule which makes it clear to whoever is the OP as to why their contribution was removed. --- ## Post 23 by @Eebzter — 2025-04-04T12:31:03Z > [@asanyan](#): > > edit: AI is also helpful in obfuscating your writing style against stylometry attacks, so prob don’t ban it Between this and assisting with a more structured wording approach, I can see exactly why it can be a useful tool. But this also requires training and understanding it is a tool for assisting not just for writing everything out for you. --- ## Post 24 by @anon80779245 — 2025-04-04T15:06:13Z I don’t think we have seen much AI posts, so it’s unnecessary IMO. --- ## Post 25 by @KevPham — 2025-04-04T15:32:19Z There was one post where a community member used Gemini to write the discussion and post because they were not confident in their Enlgish. That ended up spreading misinformation that I had to shut down really quickly. LLMs tend to try force an argument even if the parameters/evidence don’t really line up. So even if a post seems real at first, most AI arguments will exaggerate the conclusions just to follow the user prompt. --- ## Post 26 by @anon63378630 — 2025-04-04T19:39:59Z > [@KevPham](#): > > try force an argument yep, anecdotally I’ve perceived that their “personality” tries to always confirm what the user is saying, even if wrong. > [@ignoramous](#): > > is _very_ handy for CS topics […] students of my course tell me but how would they know what to look out for to truly judge it if they don’t know the subject at hand? --- ## Post 27 by @ignoramous — 2025-04-04T20:03:20Z > [@anon63378630](#): > > how would they know what to look out for to truly judge Attachments (think self-contained pedagogical material like text books and not blog posts or research papers) ground these LLMs really well. Especially if you force them to always answer with references (Gemini 2.x has a “double check” mode which they’ve found to be useful). > [@anon63378630](#): > > truly judge it if they don’t know the subject They won’t be able to if they weren’t paying attention during the lectures. I suspect the students who are making the most of it use these LLMs as an assistant to reduce time/effort spent looking things up (information retrieval) like one would, say, by searching the web or searching inside the book. Summaries and explanations may or may not always be helpful for beginners (I’ve personally found summarisation & explanation capabilities of Claude Sonnet to be quite useful for code-base exploration and other code-related tasks on _new_ repositories). For education, LLMs have the potential (mostly due to the thoroughness of the text books) but like you suspect, the killer product interface isn’t chat, and a _lot_ hinges on document understanding (tables, images, equations, diagrams, etc). --- ## Post 28 by @Eebzter — 2025-04-05T10:12:01Z I can agree that user prompt tends to be biased which results in a leaning opinion that may not even be true. I have found LLM to be inaccurate on several occasions and can even be prompted to provide incorrect answers if you want to be malicious. I like the idea of composing a better written statement or message but it can go sideways quite quickly. As a check for grammer and sentence structure it has been useful but at the same time also becomes a crutch for some. --- ## Post 29 by @win11.shading291 — 2025-04-05T15:59:42Z > [@anon42475305](#): > > Basically I’ve recently noticed an uptick in the amount of ‘AI’ generated content and I cannot recall a single time that content has genuinely been useful. In almost every such situation a human written response would not only have been more helpful, but also would’ve contributed far more positively to the forum as a whole. I seem alone in this, but I respectfully disagree. AI helped in this situation for example and helped the discussion in that thread: > [@Best tool to disable telemetry on Win11](https://discuss.privacyguides.net/t/best-tool-to-disable-telemetry-on-win11/18516/43): > > Ok, so I had to do these steps to make the Diagnostic Data Viewer Windows tool to work: 1- Settings \> Privacy & Security \> Diagnostic and feedback \> Send optional Diagnostic Data ON 2- Settings \> Privacy & Security \> App Diagnostics \> App Diagnostic access ON 3- Run \> Services.msc \> Connected User Experiences And Telemetry \> Properties \> Statup type \> Automatic 4a- CMD \> Dism /Online /Cleanup-Image /CheckHealth 4b- Dism /Online /Cleanup-Image /ScanHealth 4c- Dism /Online /Cleanup-Image /Re… I wasn’t getting the answers I was looking for and AI helped in that specific case. And I know it didn’t help only me as well. I do agree that LLM generated content should be looked with extra glasses, but a blanket ban would be undesirable IMO. Also, I previously got a topic banned saying : > Use a search engine and the forum search before posting. No excuses. Ask specific questions after you did that. In which I replied: Hi, there are no topics on this…? In which I got a reply from mod: > Feed your literal question e.g. into perplexity.ai My point here being if you ever decide to go ahead with this (which I’m against for the reasons stated above), please be consequential. --- ## Post 30 by @forkedpiccalilli — 2025-04-09T12:25:44Z > [@win11.shading291](#): > > In which I got a reply from mod: You should post a screenshot so we can see which mod suggested you use AI. --- ## Post 31 by @ph00lt0 — 2025-04-09T12:33:40Z we already banned low effort posts. Any AI post faling in that category we can just remove. Posts that used AI for better structure, better grammar (like we also recommend with LanguageTool) or translation is really fine imho. This is really a none issue --- ## Post 32 by @ph00lt0 — 2025-04-09T12:37:19Z it’s cuz we actually already remove the crap. --- ## Post 33 by @anon80779245 — 2025-04-10T20:32:26Z Ah got it :sweat_smile: --- ## Post 34 by @win11.shading291 — 2025-04-11T02:10:07Z I intentionally left that part out. This wouldn’t serve anything. --- ## Post 35 by @ignoramous — 2025-04-11T14:44:23Z > [@ph00lt0](#): > > we actually already remove the crap What kind of self-congratulatory post is this? It’d be good if y’all ever _only_ removed crap. When questioned about the removals, all I’ve seen is either stonewalling or more crap from y’all. Be humble. > [@ph00lt0](#): > > Posts that used AI for better structure You are mudding the topic which is “AI generated” content and not translation or better structure or any of the other tangential things. In general, AI slop is hard to read and may result in proliferation of high effort posts that are actually low effort. A _ban_ on users / bots if they obsessively post _gen AI_ content (without being explicit that they are) is a valid concern, and not a “non issue”. The main problem as a human user I see is, dreading a discussion with AI content, no matter if it is a bot or another user using a bot. --- ## Post 36 by @KevPham — 2025-04-11T16:10:13Z I think those concerns are completely valid. Although a lot of staff members have varying opinions on this topic, I do personally support some formal implementation of this rule to inform new members. Generally though, if you see anything that is obviously AI generated, feel free to flag it it as we will remove it. Most moderators (besides me :/) are volunteers so removal won’t always be timely. --- ## Post 37 by @Niek-de-Wilde — 2025-04-11T18:46:35Z > [@ignoramous](#): > > Be humble Something something pot and kettle. --- ## Post 38 by @jonah — 2025-04-11T20:07:13Z So, it’s sounding like we want to add something like this to the [FAQ](https://discuss.privacyguides.net/faq)? > ## Post Only Your Own Thoughts > > Using AI to generate post content is forbidden. It is always better to leave a topic unanswered, than to reply with anything other than your own original thoughts and research. > > Privacy Guides currently has no stance on AI as a research tool. However, if a poster wanted to know how AI would solve their problem or answer their question, they could ask a variety of commercial or self-hosted AI tools instead of asking here. Therefore, it is already implied that any AI-generated answer is automatically not useful to the discussion. > > This website is commonly cited as a source by AI tools when people use AI to research privacy-related topics. If we allowed AI-generated content, then this research could become self-referential and reinforce inaccurate claims. This forum is not meant to be a personal help desk where every single question gets instantly answered, it is meant to be a collection of high-quality knowledge about privacy and security. * * * **Edit:** Completed in [AI generated content is forbidden](https://discuss.privacyguides.net/t/ai-generated-content-is-forbidden/26625/1) --- ## Post 39 by @anon42475305 — 2025-04-11T21:09:38Z I am of course very interested in hearing others opinions but personally I am very happy with this as the solution so for the time being I’ll mark it as such. I think you’ve done a really great job of covering all the main concerns with generative AI and fitting it with the style of the rest of the FAQ.