Wikipedia bans ai from writing articles to stop knowledge rot

Wikipedia’s volunteer editors have drawn a hard line: no machine will author humanity’s shared encyclopedia. By an overwhelming vote they rewrote the rules last week, barring generative ai from creating, expanding or editing articles in any language edition.

The move is less a Luddite reflex than triage. For two decades the site banked on human curiosity; lately it has been flooded with semi-literate auto-text that reads like a confident freshman who never opened the textbook. Editors call it “ai slop”—the same protein-grey paste that already clogs YouTube explainers and TikTok factoids—and they want it off the lawn.

The fine print still lets robots spell-check

Grammar bots and translation helpers survive, but only if they stay dumb. Suggest a comma, fine. Insert a paragraph of synthetic prose, instant revert. Translators must certify they speak the source language; no Google-Translate-and-pray.

The clampdown arrives as scraping armies keep hammering Wikipedia’s servers, downloading terabytes to train the very models that now vomit derivative entries back onto the site. Engineers brokered a side deal: tech giants can sip a packaged, rate-limited feed instead of DDoS-ing the donation-funded boxes. Call it a paid fire escape from the fire they helped start.

Britannica is already in court accusing OpenAI of xeroxing its pages; Wikipedia prefers preventive surgery. Whether the scalpel holds depends on 200,000 volunteers who police recent changes one diff at a time. The bots never sleep, but neither, for now, do the humans who still care about footnotes that actually exist.