--- title: Wikipedia's Cleanup of AI-Generated Content | daily.dev description: Wikipedia's volunteer WikiProject AI Cleanup is tracking a slow decline in AI-generated submissions after the site banned editors from using LLMs to write... canonical: https://daily.dev/posts/wikipedia-s-cleanup-of-ai-generated-content-ok4jufv4y twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Wikipedia's Cleanup of AI-Generated Content | daily.dev og:description: Wikipedia's volunteer WikiProject AI Cleanup is tracking a slow decline in AI-generated submissions after the site banned editors from using LLMs to write... og:url: https://daily.dev/posts/wikipedia-s-cleanup-of-ai-generated-content-ok4jufv4y og:image: https://api.daily.dev/og/posts/Ok4jufV4Y.png og:image:alt: Wikipedia's Cleanup of AI-Generated Content og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Wikipedia's Cleanup of AI-Generated Content **[Hacker News](https://daily.dev/sources/hn)** · 6 min read · 1 upvotes · 0 comments ## Summary Wikipedia's volunteer WikiProject AI Cleanup is tracking a slow decline in AI-generated submissions after the site banned editors from using LLMs to write articles. Using the detection tool Pangram, founding member Ellia Lebleu estimates AI flagged about 20 percent of new articles and drafts in late 2025, dropping to 15 or 16 percent in early 2026. It's unclear whether that drop comes from the new policy or from AI models getting harder to detect. Fake or misleading citations remain common, and new tools like an AI source verifier and the Wikimedia Foundation's Paste Check aim to catch AI-written or AI-pasted text. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What percentage of new Wikipedia articles get flagged as AI-generated? Around 15 to 16 percent of new articles and drafts were flagged as AI-generated in early 2026, down from about 20 percent in late 2025, according to estimates using the detection tool Pangram. It's unclear whether the drop reflects the effect of Wikipedia's new policy banning LLM-written articles or models becoming harder for Pangram to detect. Pangram showed no false positives when tested on pre-ChatGPT Wikipedia content. _Anyone tracking how AI content detection rates shift over time can follow coverage like this on daily.dev._ ### Does Wikipedia allow editors to use AI like ChatGPT to write articles? No, English Wikipedia prohibited editors from using large language models to generate articles. Despite the rule, AI-generated content still gets submitted, mostly by beginner editors unfamiliar with the policy. Wikipedia has responded by making the rule more visible in welcome messages, editing help, and advice for new contributors. _Developers weighing AI-generated content policies for their own projects can track these debates on daily.dev._ ### How does Wikipedia detect AI-written text and fake citations in articles? Editors look for structural patterns like the 'rule of three' and negative parallelisms, plus behavioral signals such as someone producing multiple thousand-word articles within minutes. For citations, AI often finds a real source but misrepresents its content rather than inventing a fake one, a problem described as extremely common. New tools, an AI source verifier and the Wikimedia Foundation's Paste Check, help flag suspect text and sources. _Those building AI detection or content verification tools can follow real-world techniques like these on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 2 discussions and 141 comments across hackernews (as of 2026-10-27). **TL;DR:** Commenters largely support Wikipedia's push to curb AI-generated submissions, seeing LLM text as fluent but hollow and prone to fake citations, though many question whether AI content is inherently worse than human-written slop and worry about deeper issues like bias, training-data feedback loops, and verification tooling. **Sentiment:** 25% positive · 45% mixed · 30% skeptical **The case for** - Cleanup efforts are broadly welcomed as necessary since AI content tends to get noisy and generic at scale. - Some argue AI can be genuinely useful for turning rough drafts into more readable prose or assisting with research and fact-checking. - A few note that if AI-generated content is actually correct, its origin shouldn't matter as much as its accuracy. **The pushback** - LLMs fundamentally can't track provenance of facts, so they can't reliably cite or verify sources the way academic writing does. - There's concern about a feedback loop where AI trained on Wikipedia pollutes Wikipedia, degrading future training data, especially for low-resource languages. - Several worry that AI-written or algorithmically 'improved' text sounds smooth but is subtly wrong or soulless compared to human writing. - Commenters flagged Grokipedia as a cautionary example of AI-generated reference content being unreliable and ideologically skewed. - Debate over whether Wikipedia's own content already suffers from political/ideological bias independent of AI, complicating the cleanup narrative. **By community** - hackernews (mixed): Support for AI cleanup is tempered by long tangential debates over bias, provenance, writing quality, and whether AI-written content differs meaningfully from existing human-written issues. **Hottest debate:** Whether Wikipedia (and encyclopedic content generally) already has ideological bias baked in, separate from the AI issue, with examples like the HIV/AIDS article cited as evidence on multiple sides. **Open questions** - Can LLMs ever be made to reliably track and cite the provenance of factual claims? - Would AI companies have an incentive to fund or support detection/verification tools like Pangram to protect their own training data quality? - Is the perceived decline in Wikipedia's writing quality actually attributable to AI, or a pre-existing trend in article bloat and editorial style? **Highlights** > LLMs cook up new sentences based on probability, based on everything in their training and their context. There's just nothing in their model that traces provenance of facts, they physically don't store that information. Once they have written a sentence, they can't tell you from which work they got this information. There are tools where they try to retroactively add references, the LLM interfaces try to do this, but it's usually inaccurate. I think of the way a scientific article works. "The sky is blue (Pearson, 1992)." The writer is fully aware what the factual statement is and which publication supports it (not just that, but which kind of publication - primary, secondary source, peer reviewed?). An LLM can try to imitate this, but the actual information to support this does not exist in its training. > — [kzrdude on hackernews](https://news.ycombinator.com/item?id=50258554) > A little over a year ago, the MIT Technology Review ran a piece about how outputting AI content to non-English Wikipedia is rapidly corrupting the primary training data for uncommon languages because of exactly this process: https://www.technologyreview.com/2025/09/25/1124005/ai-wikip... > — [DonaldPShimoda on hackernews · 2 comments](https://news.ycombinator.com/item?id=50259720) > Even putting aside accuracy, if the plan is for the tools that verify the written content to be in any shape or form outside of the human loop, then this feels like it glosses over the issues of both passive bias in data & active bias (e.g data filtering etc. ) when training such models that you'd be poisoning this information sources with. Even subtle bias can manifest in significant ways. Given that the predominant market leaders in producing such models are currently : A country that's attempting to squeeze the word "america" in any geographic formation possible + rewriting history to put itself first, and a country that has perfected censorship en-mass and already actively doctors its own history. I'm not convinced giving LLMs carte blanche to write authoritative sources on topics that - often are ideologically sensitive/controversial isn't just needlessly reckless and short-sighted. It's not a far reach to imagine such models reflecting the current views of administrations in these countries and becoming reluctant to mention certain colonial ambitions, environmental damage, political corruption, etc. where it might be relevant to authoritative content. Wikipedia already struggles with issues in idealogical bias from authors - human validation (somewhat)helps mitigate that. You could argue that human validation is still prone to same bias - but you can always draw on a more diverse pool of editors with little ideological overlap. With LLM validation you're pretty much at the mercy of a pretty incestuous pool of data & a handful of self-serving corporates & governments. > — [steampoweredslo on hackernews](https://news.ycombinator.com/item?id=50258535) > Much of wikipedia is written by one person, Steven Pruitt. He has created more than 33,000 Wikipedia articles. People rightly credit him for cataloging obscure topics even though much of what he writes is rote. But now when AI does it, it's deemed bad? If it's factual and at all useful to someone, why not leave it? I feel like we're waging the same battles when Wikipedia became an open encyclopedia. Surely you need to control who can edit right? We all know how that ended. https://en.wikipedia.org/wiki/Steven_Pruitt > — [bko on hackernews · 4 comments](https://news.ycombinator.com/item?id=50258452) > I attempted to fix some AI-generated content that I found on wikipedia recently - it had broken formatting, confusing + made-up text, and links on the wrong keywords; just generally very low quality edits. I made my edits and they were immediately reverted by wikipedia's own AI bot. I gave up. > — [liveoneggs on hackernews · 3 comments](https://news.ycombinator.com/item?id=50260225) **Source threads** - [hackernews](https://news.ycombinator.com/item?id=50257326) · 271 points · 139 comments - [hackernews](https://news.ycombinator.com/item?id=50182203) · 5 points · 2 comments ## Similar posts on daily.dev - [Generative AI and Wikipedia editing: What we learned in 2025](https://daily.dev/posts/generative-ai-and-wikipedia-editing-what-we-learned-in-2025-fxcbzqmcr) · Hacker News · 0 upvotes · 0 comments - [ArXiv introduces one-year ban for researchers who submit papers with unchecked AI-generated content](https://daily.dev/posts/arxiv-introduces-one-year-ban-for-researchers-who-submit-papers-with-unchecked-ai-generated-content-ql5uy3yuq) · The Next Web · 2 upvotes · 0 comments - [The best guide to spotting AI writing comes from Wikipedia](https://daily.dev/posts/the-best-guide-to-spotting-ai-writing-comes-from-wikipedia-sbx4xvgmp) · TechCrunch · 1 upvotes · 1 comments - [Wikipedia:WikiProject AI Cleanup](https://daily.dev/posts/wikipedia-wikiproject-ai-cleanup-1g2nqa8sb) · Hacker News · 1 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm) [View this post on daily.dev](https://daily.dev/posts/wikipedia-s-cleanup-of-ai-generated-content-ok4jufv4y) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Wikipedia's Cleanup of AI-Generated Content","url":"https://daily.dev/posts/wikipedia-s-cleanup-of-ai-generated-content-ok4jufv4y","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/wikipedia-s-cleanup-of-ai-generated-content-ok4jufv4y"},"datePublished":"2026-10-27T14:26:02.204Z","dateModified":"2026-10-27T17:27:26.249Z","description":"Wikipedia's volunteer WikiProject AI Cleanup is tracking a slow decline in AI-generated submissions after the site banned editors from using LLMs to write...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/5284b316482b916332ee8429ef7a5d7e?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/5284b316482b916332ee8429ef7a5d7e?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Hacker News","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Hacker News","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/hn","url":"https://daily.dev/sources/hn"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/wikipedia-s-cleanup-of-ai-generated-content-ok4jufv4y","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hacker News","item":"https://daily.dev/sources/hn"},{"@type":"ListItem","position":3,"name":"Wikipedia's Cleanup of AI-Generated Content"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/wikipedia-s-cleanup-of-ai-generated-content-ok4jufv4y#faq","mainEntity":[{"@type":"Question","name":"What percentage of new Wikipedia articles get flagged as AI-generated?","acceptedAnswer":{"@type":"Answer","text":"Around 15 to 16 percent of new articles and drafts were flagged as AI-generated in early 2026, down from about 20 percent in late 2025, according to estimates using the detection tool Pangram. It's unclear whether the drop reflects the effect of Wikipedia's new policy banning LLM-written articles or models becoming harder for Pangram to detect. Pangram showed no false positives when tested on pre-ChatGPT Wikipedia content. Anyone tracking how AI content detection rates shift over time can follow coverage like this on daily.dev."}},{"@type":"Question","name":"Does Wikipedia allow editors to use AI like ChatGPT to write articles?","acceptedAnswer":{"@type":"Answer","text":"No, English Wikipedia prohibited editors from using large language models to generate articles. Despite the rule, AI-generated content still gets submitted, mostly by beginner editors unfamiliar with the policy. Wikipedia has responded by making the rule more visible in welcome messages, editing help, and advice for new contributors. Developers weighing AI-generated content policies for their own projects can track these debates on daily.dev."}},{"@type":"Question","name":"How does Wikipedia detect AI-written text and fake citations in articles?","acceptedAnswer":{"@type":"Answer","text":"Editors look for structural patterns like the 'rule of three' and negative parallelisms, plus behavioral signals such as someone producing multiple thousand-word articles within minutes. For citations, AI often finds a real source but misrepresents its content rather than inventing a fake one, a problem described as extremely common. New tools, an AI source verifier and the Wikimedia Foundation's Paste Check, help flag suspect text and sources. Those building AI detection or content verification tools can follow real-world techniques like these on daily.dev."}}]} ```