Wikipedia completely forbids articles created by neural networks

New Wikipedia Policy: Ban on Using Neural Networks for Editing

At the end of last week, Wikipedia leadership updated its rules by introducing a restriction on writing and rewriting articles using neural networks. The text of the innovation states that articles created with AI often violate “several key content rules” of the platform.

Who it Applies To

Currently the rule applies only to the English-language version of Wikipedia. However editors can still use large language models (LLM) in certain cases:

- Preliminary proofreading: LLM may suggest basic edits but should not add new content.
- Translation: AI can be used to translate articles from other languages into English, provided accuracy requirements and sufficient knowledge of the source language are met.

How to Determine Acceptability

The guide emphasizes that some authors’ style may resemble that of an LLM. Therefore editors must rely not only on stylistic cues but also on compliance with core content rules. It is recommended to consider:

1. Whether the text meets Wikipedia’s key requirements.
2. Recent edits by that author.

Community Reaction

The discussion of the innovation lasted several days, but ultimately the vote ended in a decisive majority for adopting the policy. Editors noted that the rule targets “explicitly problematic” cases of LLM use while leaving room for acceptable technology application.

Context of Fighting AI Content

Wikipedia has been dealing with the problem of articles entirely written by AI for months. In response, the community developed several measures:

- A “quick deletion” policy for low‑quality articles.
- WikiProject AI Cleanup – an initiative to identify and remove content created by neural networks.

25 Years of Wikipedia

On January 15, Wikipedia celebrated its 25th anniversary. Over that time the site grew from a small project with a few hundred pages into the world’s largest knowledge source: more than 65 million articles in various languages and about 15 million monthly views.