Overview
- Volunteer editors approved the policy in a Request for Comment that closed on March 20 with overwhelming support.
- The rule bars using large language models to write or rewrite articles, with narrow exceptions for basic copyedits to an editor’s own text and first‑pass translations that a fluent human must verify.
- Editors adopted the ban because AI systems often hallucinate facts, fabricate or mangle citations, and subtly change meaning beyond what sources support, all of which conflict with Wikipedia’s core content policies.
- Enforcement relies on human moderation since AI‑detection tools are unreliable, and repeat misuse can be treated as disruptive editing that may lead to blocks under normal appeal processes.
- The guideline applies only to English Wikipedia, while other language editions set their own rules, with Spanish Wikipedia already enforcing a stricter ban, and the change aims to curb a feedback loop where flawed AI text can be scraped and fed back into future models.