Wikipedia’s volunteer community has moved from cautiously experimenting with generative AI to sharply limiting it. By March 2026, editors found that AI-written material violated the encyclopedia’s core principles often enough to ban large language models from article writing, leaving translation and minor proofreading as permitted uses.
The warning signs appeared earlier. A Princeton study found that about 5% of nearly 3,000 English Wikipedia articles created in August 2024 contained substantial AI-generated text. Editors encountered generic prose, fabricated sources and even remnants of prompts left in submissions, creating more verification and cleanup work for volunteers.
At the same time, AI summaries in search engines are reducing the need for people to visit Wikipedia directly. The encyclopedia still supplies information to those systems, but readers may see its work only through an AI intermediary. That weakens the connection between public use and the human contributors who maintain the underlying source.
The risk is a feedback loop: fewer readers and contributors leave less capacity to correct weak automated content, while AI products continue relying on Wikipedia as a reference. The community’s ban protects editorial control, but it cannot by itself solve declining traffic or ensure that downstream AI systems preserve the context and uncertainty in human-edited articles.