Wikipedia has implemented a sweeping policy banning the use of large language models to generate or rewrite encyclopedia articles. The move responds to growing concerns over AI-generated hallucinations, fabricated citations, and erosion of verifiability-the bedrock of Wikipedia’s credibility.
Editors may still use AI for basic copyediting-such as grammar or spelling corrections-provided no new content is introduced and all changes are manually reviewed.
"Text generated by large language models often violates several of Wikipedia's core content policies," the update states. "There is no accountability, no connection to what someone believes or stands behind," said Dr. Emily M. Bender, a linguistics professor at the University of Washington. "That’s not there for large language models."
The policy does not impose automatic penalties but warns that repeated misuse constitutes "disruptive editing," potentially leading to account blocks. Appeals are possible through administrative review or the Arbitration Committee.
Joseph Reagle of Northeastern University noted that many AI models are trained on Wikipedia data, creating tension as tech giants like Microsoft, Google, Amazon, and Meta license its content commercially. "There's antipathy toward services that appropriate community work then flood it with AI slop," he said.
AI-assisted translation between language editions remains permitted, but only after editors verify the original text. The policy cautions against relying on stylistic cues alone to detect AI output-compliance with sourcing rules, not writing patterns, is the decisive factor.
