[00:20:47] In this specific instance, if raidarr is right and the original article is AI generated, then it really isn't a copyright violation and simply spam, at least under U.S. law. [Anything AI generated isn't copyrightable under U.S. law.]() [01:28:19] [1/3] He assumes it is AI generated. As there is already a document online, that is material, published under copyright law. It is for Raidarr to prove that it is AI generated. [01:28:20] [2/3] AI generated is indeed not copyrighted in the US (Nor in Europe for that matter) [01:28:20] [3/3] But we cannot assume something is AI generated by looking at a document and see it is large. [01:30:10] [1/2] And there is a very basic difference here as well. [01:30:10] [2/2] Meta is an official Miraheze Wiki. Miraheze is responsible for what it hosts. What personal wikis do, is a different matter [01:38:57] [1/7] Let's examine the article deeper: [01:38:57] [2/7] - Oddly consistent title format - a phrase that's usually followed by a colon and a statement stating the actual focus of the paragraph [01:38:58] [3/7] - Excessive use of dashes - while em dashes aren't a sign of AI writing by themselves, the probability of the entire work being AI-generated increases with each dash [01:38:58] [4/7] - Excessive use of bolding to highlight key phrases - the only other place I've seen bold markup used this much is on the reception wikis [01:38:58] [5/7] - Excessive use of variations of the "not just X but Y" sentence structure, which is universal in AI writing [01:38:58] [6/7] - Items in a list are preceded by a word capturing the essence of that item more often than expected [01:38:59] [7/7] If you plug a sample paragraph of the essay into a few AI content detectors, I can say that most if not all of them will say that it's AI-generated with 99.9% confidence; therefore, I would assume that this is in fact an AI-generated essay, which means that it cannot be copyrighted, ergo there is no copyright violation [01:50:37] (Me as a genuine em dash user: 💔 ) [01:52:36] [1/6] Countering: [01:52:37] [2/6] Reasons you cannot reliably conclude it’s AI-generated [01:52:37] [3/6] * Current AI-detection tools are not reliable: they often produce false positives and cannot definitively prove AI origin. [01:52:37] [4/6] * Even human-written text can exhibit machine-like features such as uniform sentence length, high formality, repeated phrasing - which are sometimes flagged by AI detectors but are not proof of AI origin. [01:52:38] [5/6] * The author may have used an AI for rewriting, polishing and summarising the article, even if the core idea is human-generated, making attribution ambiguous. [01:52:38] [6/6] It's possible. Highly likely even. But in academics, this is generally a style in which a paper is written in. [01:56:10] [1/4] - Not 100%, but they're reasonably accurate a lot of the time [01:56:11] [2/4] - Not as much as AI tends to exhibit those factors [01:56:11] [3/4] - Some AI detectors can also detect when a certain text is written by humans and polished by AI [01:56:11] [4/4] Not only is the essay being AI-generated entirely possible, it's the only reasonable conclusion upon reading it - humans, even when using the highest level of formality, do not write like this [02:01:53] (genuine question and not meant to be an insult or accusation in any shape or form, did you use AI to write this message?) [02:07:08] [1/2] * Should we really waste time debating whether an article is, or isn't AI-generated, or a copyright violation? — I think we do. I have always stated that we should be cautious allowing people to post random things on Meta. I rather have users requesting a wiki for that purpose. Why take the risks of hosting content that might be copyrighted? [02:07:09] [2/2] * I always find it difficult that people want to walk close to the edge of a slippery slope. I rather keep myself a safe distance from the edge. [02:07:13] [1/2] I wasn't really angling to argue the ai point so much as attempting to make it clear that our global functionary job on copyright is not to be proactive police and cast proactive judgement on the subject, but to take action to legitimate reports from rights holders which is the practical obligation of the farm beyond which it is unreasonable for miraheze to take on. Unless of course [02:07:13] [2/2] it is a particularly obvious well known instance [02:07:51] in this case I want to see what the ostensibly bot account has on its script right now but if it does nothing shortly it's going to get deleted regardless because it's assuredly spam [02:07:59] Pisses me off that dashes are treated like AI stink. I’ve been using dashes since before OpenAI even existed [02:08:44] It's from Wikipedia and Google search AI that gives an overview on reliability on AI [02:10:01] It's the overuse of em dashes that gives away AI writing, not the use of em dashes themselves - that could've been made clearer [02:10:37] the obviousness is much clearer when you just look at the account that posted it tbh [02:10:42] but again that's really a secondaryp ointer [02:12:50] [1/2] the article exhibits every common trait of ai so I'd stand by that confidently too but that is why standard procedure on most of the internet is that the creator makes the claim and the platform responds, this is how the internet functions with the bedrock of section 230 and why it does not behoove a platform to try and mark out every possible case, a standard for which wikipedia is [02:12:50] [2/2] extremely unusual [02:13:09] last 2c on me for that for the night [02:13:15] [1/2] The original poster https://substack.com/@johnobidi (which posted the article on https://medium.com/@jonimisiobidi/the-theory-of-entropicity-toe-a-new-path-toward-the-unification-of-physics-e4e2ac3857a8 ) writes about himself: Love of physics and science, philosophy and literature. [02:13:15] [2/2] It isn't proof that it is AI written, nor does it proof it wasn't. But the likelihood such a person writes in a highly formal way is rather high, in academic circles. [02:14:10] [1/2] this is an ai account. [02:14:10] [2/2] https://cdn.discordapp.com/attachments/443926951292567562/1433277772016652308/image.png?ex=69041b71&is=6902c9f1&hm=2fdd897a0e18f56bd92189a32e4e5f9a85471b5519bd4f2477fc96541d27600c& [02:14:21] you can recognize it or not, I have seen many tells and can tell you with certainty it is, and we can agree to disagree if you so desire [02:15:23] I think this discussion has gone circular [02:17:07] [1/2] The website linked in the profile is literally about AI strategy, which should be the final nail in the coffin for this dispute; I will not make further comments [02:17:08] [2/2] https://cdn.discordapp.com/attachments/443926951292567562/1433278517172502549/image.png?ex=69041c23&is=6902caa3&hm=09c20f2caf0aedaec760b80cf17937dd96ededed8b89698320a1e61e6314e9ad& [02:29:02] [1/4] We wouldn't even have these conversations, including AI generated text, Spam, copyvio, etc. if we would use Meta within the scope of the wiki. [02:29:02] [2/4] The CP itself is clear that a wiki needs to stay within the scope of the wiki. [02:29:02] [3/4] As Meta is a central organisation wiki for Miraheze, things would just be easier to deal with when there is no content that has nothing to do with Miraheze or the user. All encyclopaedic pages, contest tables, and other nonsense would be removed. That would be nice. [02:29:03] [4/4] Oh well — I keep dreaming of a better time, where more clear rules are introduced. [02:30:14] Individual note for posterity: this conversation started because of a page on Login Wiki, not Meta - if it were posted on Meta, it probably would've been deleted by a Meta admin [02:33:57] [1/4] True. But we do get this on both Login and Meta. I find it useful to discuss this, eventhough we do not always agree. [02:33:58] [2/4] I ran it through the AI: [02:33:58] [3/4] ```Based on external AI-analysis tools (e.g., GPTZero, Writer.com, Copyleaks), text segments from that article show 70–85% probability of AI involvement — mainly due to its uniform sentence rhythm, low lexical entropy, and overuse of explanatory parallelism. [02:33:58] [4/4] So: likely AI-assisted, though not certainly fully AI-generated — it reads as a hybrid of human conceptual framing with AI-polished prose.``` [04:25:19] [1/2] Late to the party, but for any latecomers who feel like engaging, going to ask that we call it a night on this. [04:25:20] [2/2] Argument has gone circular as others noted and meta admins are approaching a draft guidance ready for input. [11:22:37] [1/2] https://meta.miraheze.org/wiki/User_talk:Rodejong?diff=prev&oldid=498528 [11:22:37] [2/2] @CVT [11:24:17] Reverted, and blocked. [11:24:35] cheers <3 [11:25:34] Also, from the looks of things, I suspect that account is most likely a repeat-customer based on the hostility. [11:27:41] [[Special:Diff/498509]] [[Special:Diff/498527]] Aha, looks like it to me. [11:27:41] https://meta.miraheze.org/wiki/Special:Diff/498509 https://meta.miraheze.org/wiki/Special:Diff/498527 [11:27:46] sure looks like it, perhaps our LTA fans would know [11:27:48]

[1/2] [11:27:49]

[2/2] [11:28:14] I like how they tried to get GA [11:28:37] @Stewards ^