Wikipedia’s English site has enacted a policy banning AI-generated text in article content, with two exceptions for editing refinement and translation aids after thorough review.

Editors are cautioned that large language models can alter meaning or misrepresent sources if not carefully reviewed.

To comply, editors can use AI to refine writing without changing factual content, and for translations, AI can draft a first pass but editors must verify sentence by sentence for accuracy.

The move is framed as part of a broader cultural shift and resistance to rapid AI expansion across online communities.

The policy reflects ongoing debate about LLMs and the challenges of guidelines, with administrator Chaotic Enby noting consensus favored change but implementation remains complex.

The decision sits amid concerns about AI accuracy, hallucinations, and balancing rapid AI content with human-verified knowledge.

The policy aims to preserve verifiability, sourcing, and overall accuracy in a community-driven knowledge platform.

A Wikipedia administrator argues for broader, community-driven consideration of AI use and views the policy as a response to AI-driven platform changes.

The same administrator describes the policy as a grassroots pushback against AI’s influence and platform enshittification, highlighting wider community considerations.

Notes reference tools like ChatGPT and Google Gemini, with a disclosure about a lawsuit involving OpenAI’s training data.

The stance follows extensive debate and reflects a cautious approach that prioritizes human oversight over AI-generated content.

The move reflects a broader issue for knowledge platforms: using AI to support rather than replace human editors, preserving content quality and verifiability.