The English-language section of Wikipedia has tightened its rules on the use of large language models: editors are now explicitly prohibited from creating or rewriting articles using AI. However, a complete ban on such tools has not been imposed.
The English-language section of Wikipedia has tightened its rules on the use of large language models: editors are now explicitly prohibited from creating or rewriting articles using AI. However, a complete ban on such tools has not been imposed.
The new rule, as outlined on Wikipedia:Writing articles with large language models, states that using LLMs to generate or rewrite articles is prohibited. A shorter version of the document states this even more simply: «Do not use large language models to generate article content.» The community explains this by saying that texts generated by such systems often violate basic principles of encyclopedias, including requirements for verifiability, accuracy, and relevance to sources.
At the same time, Wikipedia is not abandoning AI completely. Editors are allowed to use language models for basic edits of their own text after human verification, but only on the condition that the system does not add new content. Separately, the rules warn: even when the editor requests only stylistic editing, the model can change the meaning of a phrase so that it no longer corresponds to the specified sources. There is also a separate instruction on LLM-assisted translation for translating articles from other language sections.
The new wording seems stricter than the previous one. Previously, the English-language Wikipedia primarily stated that LLM should not be used to write new articles «from scratch.» Now the ban directly covers both the creation and rewriting of encyclopedic content.
For Wikipedia, this is not just a technical clarification, but an attempt to protect the main value of the project — text that can be checked against sources and tracked editorially. Against the backdrop of the rapid spread of AI tools, the encyclopedia leaves a key role in writing articles to humans, and assigns language models only a supporting function — minor editing and individual translation tasks under the control of the editor.
Previously, dev.ua wrote about how Wikipedia released a set of structured data specifically optimized for training artificial intelligence models. Thus, the platform is trying to discourage AI developers from using bots to collect information that burdens its servers.
