Wikipedia has spent two decades becoming the gold standard for the internet’s collective knowledge. Now, it is fighting to ensure that gold isn’t turned into lead by the weight of generative artificial intelligence (AI).
The free online encyclopedia, written and maintained by a community of volunteers, officially updated its editing policies to prohibit the use of Large Language Models (LLMs) for generating or rewriting article content. The new mandate marks a significant escalation in the platform’s battle against AI hallucinations and underscores a commitment to human-led scholarship amid a rapidly shifting digital landscape.
According to reports from 404 Media, a formal vote by editors signaled overwhelming support for the ban. The consensus reflects a growing anxiety that generative AI, while efficient, fundamentally clashes with Wikipedia’s core pillars: neutrality, verifiability, and the prohibition of original research.
“For this reason, the use of LLMs to generate or rewrite article content is prohibited,” the updated editor advice page said. The platform warned that tools like ChatGPT, Claude, and Google Gemini often produce text that “violates several of Wikipedia’s core content policies,” specifically by fabricating citations or distorting facts in a way that is not supported by source material.
Despite the sweeping ban, Wikipedia has carved out two narrow exceptions. AI may still be used for basic copyediting of an editor’s own writing and for translating articles from other language branches into the English version. However, these exceptions come with rigorous stipulations. For translations, editors must be proficient in both languages to manually verify the accuracy of the AI’s output against original sources. For copyedits, the AI is strictly forbidden from introducing any new information.
“Caution is required, because LLMs can go beyond what you ask of them and change the meaning of the text,” the policy warned, noting that human judgment must remain the final arbiter of any change.
The move comes at a time of vulnerability for the encyclopedia. As tech giants embed AI directly into search results and email suggestions, Wikipedia’s traffic is beginning to wane. Between March and August 2025, human page views dropped 8%, a sign that the paradigm shift in how people consume information is well underway.
Last year, a pilot program for AI-generated summaries was scrapped after editors argued it compromised the site’s integrity. Founder Jimmy Wales has echoed this sentiment, suggesting that while AI might help with backend chores, it is not yet capable of the nuanced, fact-based writing required for an encyclopedia.
Wikipedia’s action comes after Encyclopedia Britannica and its subsidiary, Merriam-Webster, filed a copyright-infringement lawsuit against OpenAI on the grounds its intellectual property is being “cannibalized” to fuel the development of large language models (LLMs).
Britannica’s legal team asserts OpenAI utilized nearly 100,000 copyrighted articles and dictionary entries to train its AI models without authorization. Britannica is seeking an unspecified amount in monetary damages and a court-ordered injunction to prevent OpenAI from continuing the alleged infringements.
As the internet becomes increasingly saturated with machine-written content, Wikipedia’s new policy ensures that the data these very AIs are learning from remains human-generated.
By banning the machines, Wikipedia is attempting to save the internet from becoming a hall of mirrors, ensuring that the “sum of all human knowledge” doesn’t become a mere echo of an algorithm.
