At a time when modern technologies compete to shape and guide the future of the internet, maintaining the accuracy and reliability of information remains the greatest challenge for leading knowledge platforms, especially with the wide and rapid spread of automated text generation tools. In this critical context, the English edition of Wikipedia has taken a firm and official stance by banning the use of large language models to write or rewrite articles published on its open-to-all platform.
This decision, officially adopted on March 20, marks a fundamental turning point in the free encyclopedia’s content policies. This decisive measure came after volunteer editors voted by an overwhelming majority in favor of imposing this strict restriction, in a clear effort to protect the quality of public human knowledge from informational pollution.
The new guidelines, which are now an integral part of Wikipedia’s official content policy pages, clearly state that texts generated by large language models frequently violate several of the encyclopedia’s core policies. Accordingly, it has been officially announced that using these advanced software models to create article content from scratch or rewrite it is entirely prohibited. This ban reflects a growing awareness of the risks accompanying AI-generated content, which may lack the required neutrality and fall into the trap of fabricating facts, ultimately misleading readers and severely damaging the source’s credibility.
Permitted exceptions under the ban policy
Despite the extreme strictness of this historic decision, the imposed ban is not entirely absolute. The encyclopedia’s administration has left a narrow and restricted window allowing editors to use large language models in two very limited cases, serving article quality without compromising the essence of information.
The first case involves using these technical tools to suggest basic edits and proofreading for texts written by the editors themselves, provided that the artificial intelligence does not introduce any new content or additional information on its own; its strict role is limited to improving phrasing and correcting grammatical and spelling errors.
The second case concerns translation processes, where editors are permitted to use language models to produce initial drafts of articles translated from other language editions of Wikipedia into English. In both mentioned cases, the new policy mandates a careful and comprehensive human review. Specifically in the case of translation, editors must possess a high and advanced level of fluency in the source language to ensure they catch any errors or meaning deviations the machine might introduce.
The policy clearly warned against potential risks, stating that extreme caution is required because large language models may exceed their requested tasks and intentionally alter the meaning of the original text in a way that completely contradicts what is supported by reliable and cited sources.
Combating AI-generated content pollution
This decisive step is the culmination of long and grueling months of struggle and challenges faced by Wikipedia’s active volunteer community against the rising tide and massive influx of AI-written articles. To counter this alarming phenomenon, editors previously established the “WikiProject AI Cleanup,” a dedicated and organized initiative aimed at identifying and removing machine-written content, which often suffers from serious structural and factual issues threatening the encyclopedia’s integrity.
These technical problems include what is known as AI hallucination, where artificial intelligence generates fake citations and unfounded, fabricated references. Furthermore, editors discovered stylistic markers that expose artificial, machine-generated texts, such as the excessive and unjustified use of em dashes and the stereotypical, exaggerated repetition of certain words like “furthermore” or “stunning.” These warning signs have become clear fingerprints helping volunteer teams track down non-human content and effectively cleanse the encyclopedia’s pages of it.
Accelerating deletion procedures and the future of free knowledge
To ensure the effective and conclusive implementation of this strict policy, Wikipedia administrators have been granted new and expanded powers enabling them to fast-track the deletion of AI-written articles that have not undergone review and verification by the submitting individual. This innovative new procedure bypasses the usual 7-day discussion period, allowing for immediate and decisive intervention.
Blatant signs that warrant immediate fast-track deletion include phrases directly addressing the reader that do not fit an encyclopedic tone, such as “I hope this helps!”, illogical citations, and fake references containing invalid International Standard Book Numbers (ISBNs) or broken Digital Object Identifiers (DOIs) that lead to no real source.
This proposal, submitted by the prominent community administrator known as “Chaotic Enby,” received what the community described as overwhelming support and unparalleled endorsement, reflecting public anxiety. Chaotic Enby described this new policy as an inevitable retreat movement and legitimate resistance against the phenomenon of noticeable technological degradation in internet quality, and against the forced and continuous push of artificial intelligence technologies by many tech companies over the past few years. Expressing his future aspirations, he added that his true hope is for this courageous decision to spark a broader shift in how we handle digital content to preserve intellectual integrity.
Frequently asked questions
Q: What is the new decision made by the English Wikipedia regarding artificial intelligence?
A: The English edition of Wikipedia has approved an official and strict ban on using large language models to write or rewrite articles, in order to protect content quality from errors and misinformation produced by these software models.
Q: Are editors completely barred from using AI tools on Wikipedia?
A: No, the ban is not absolute, as there are only two exceptions. AI is permitted to suggest linguistic edits and corrections on text written by the editor themselves, and to produce initial translation drafts from other Wikipedia languages, provided that careful human review is conducted in both cases.
Q: How do Wikipedia administrators verify that a presented article was written by a machine?
A: Administrators and editors rely on several warning signs, such as fake references, illogical citations, invalid International Standard Book Numbers, addressing the reader with direct phrases like “I hope this helps!”, and the excessive use of certain stereotypical words.
