English Wikipedia prohibits using generative AI to write or edit entries

Madrid. The English version of Wikipedia has banned the use of generative artificial intelligence (AI), powered by large language models (LLM), to write or edit the content of its entries, claiming that it violates its content policies, unless it is used for corrections.

The free encyclopedia platform, edited collaboratively, has specified that the use of assistants such as ChatGPT from OpenAI, Gemini from Google, Claude from Anthropic or DeepSeek, which assist writing using advanced AI capabilities, often “violate several of Wikipedia’s main content policies”, by introducing modified content by changing its meaning.

In this sense, the organization has determined that the use of these assistants, powered by large language models (LLM), to write or edit Wikipedia entries is prohibited. However, he has clarified in an entry that there are two cases in which the use of AI is allowed, to correct human errors or translate articles.

Specifically, users can use AI to review texts and correct basic errors in their own articles so that it does not introduce its own content. However, this correction process must be carried out with caution, since “LLMs can go beyond what is requested and alter the meaning of the text in such a way that it is not supported by the cited sources,” the platform has stressed.

Assisted translation

On the other hand, editors are allowed to use large language models (LLM) to translate Wikipedia entries from another language into the English version, but they must follow the guidelines indicated in the regulations for LLM-assisted translation, which require full supervision of the translation.