It was only a matter of time. Wikipedia, the enormous volunteer-run encyclopedia with more than 7.1 million articles in its English version, has officially banned the use of artificial intelligence for writing and rewriting articles. The decision came at the end of March.
The problem of AI content on Wikipedia was nothing new. Editors began noticing suspicious articles years ago. Ilyas Lebleu, a volunteer editor from France and a founding member of the WikiProject AI Cleanup group, described it this way: "We started noticing a lot of articles written in a style that we don't normally see on Wikipedia."
The community responded gradually. First, internal guidelines were created on how to identify AI content. Editors learned to look for telltale signs: fabricated citations, broken links, overblown phrases, and sudden changes in style. Then came the option to quickly delete suspicious articles. But even that was not enough. "In recent months, more and more administrative reports revolved around problems with language models, and editors were overwhelmed," Lebleu told 404 Media.
Vote on the AI ban
The proposal for the ban was submitted by Lebleu himself, who goes by the username Chaotic Enby on Wikipedia. The vote took place on March 20, and the result was decisive. Forty votes in favor, two against. The new rules were then added to Wikipedia's official guidelines.
The new policy states clearly: "Text generated by large language models (LLMs) often violates several of Wikipedia's core content policies. For this reason, using LLMs to generate or rewrite article content is prohibited." The main reasons? Inaccuracies, fabricated information, and violations of the principles of verifiability and neutrality. Chatbots simply hallucinate. And on an encyclopedia where every claim must have a source, that is a fatal problem.
The two exceptions that remain
The ban is not absolute. Wikipedia has left two avenues open for artificial intelligence.
First, editors may use AI to suggest minor language corrections in their own texts, but only if the model adds no new content and every change is reviewed by a human. However, the new rules warn: "Caution is necessary because language models may go further than you ask them to and change the meaning of the text in ways that are not supported by the cited sources."
Second, AI may help translate articles from other language versions of Wikipedia into your language. But here too, there is a condition: the editor must be proficient enough in both languages to verify the translation independently.
The irony of the whole story
The whole story has one bitter twist. Wikipedia probably helped train the very same language models that now threaten its existence. Millions of articles that volunteers spent years writing became the foundation for ChatGPT and similar tools. And the result? Last year, ChatGPT overtook Wikipedia in monthly visits. Wikipedia's traffic fell by 8% at the end of 2025 compared with 2024. Founder Jimmy Wales called it a "mess" and warned last year that, from Wikipedia's perspective, current models are "nowhere near good enough".
Lebleu sees this decision as more than just an internal rule for a single website. "I predict a domino effect that will empower communities on other platforms to decide whether they welcome AI on their own terms," he said.
He may well be right. When even Wikipedia, which effectively helped build the internet as we know it, decides to tell chatbots to stop, other platforms are sure to take notice.
Additional sources: cnet.com and theverge.com
