Wikipedia has prohibited the use of large language models to create or rewrite article content, citing violations of core content policies. Basic edits like fixing typos and certain article translations are permitted under strict conditions. The policy's enforcement details remain unclear.
Wikipedia, the open-source online encyclopedia, has updated its editing policy to ban text generated by large language models such as ChatGPT and Google Gemini. The policy states that such AI-generated content 'often violates several of Wikipedia's core content policies,' prohibiting its use to generate or rewrite articles except in limited cases. This move aims to maintain the platform's reliance on reliable, verifiable sources amid the rise of AI tools in content creation. The exceptions include very basic edits, such as correcting typos or adjusting formatting, but only for articles written by the editor themselves and after review by a Wikipedia volunteer or administrator. Even then, editors are cautioned that AI might alter meanings inaccurately. Full-article translations from other language versions of Wikipedia into English are also allowed, provided the translator is fluent in both languages and adheres to all policies. Wikipedia's editing policy page reflects this change, though the exact implementation date is unclear, as a representative did not respond to inquiries. Last year, the Wikimedia Foundation urged AI companies to cease scraping Wikipedia data and instead use its Enterprise API to access content sustainably while supporting the nonprofit's mission. No specifics on enforcement or penalties for violations appear in the policy. This decision underscores ongoing concerns about AI accuracy, errors, and plagiarism, even as such tools integrate into smartphones and services like Apple Intelligence and Galaxy AI.