Wikipedia bans AI-written content in articles

Wikipedia, the free encyclopedia that anyone can edit, has announced that ‘anyone’ only includes humans and not AI. Specifically, Wikipedia has updated its guidelines to prohibit the use of large language models to generate or rewrite article content on its site.

Allowing anyone to edit and contribute to Wikipedia has allowed it to flourish, as experts on any subject matter can play a part in sharing their knowledge to others. However, the rise of large language models and generative AI such as ChatGPT and Gemini has meant that some contributors are beginning to rely on them to generate article text.

LLMs are at their core pattern completion programs. Their training data can often times include fiction, forums and generally low quality content, leading the LLM to draw conclusions that are not actually there in any reliable source. Another major issue is hallucination, where LLMs conjure up statements not backed by anything credible, which Wikipedia calls original research at best, and outright fabrication at worst.

The only few exceptions where Wikipedia allows LLMs to be used is when editors want them to help suggest basic copyedits to their contribution. They still ask for caution when using it this way though, as LLMs may end up changing the meaning of the text you provide it. LLMs can also be used to translate articles, as long as you adhere to their LLM-assisted translations guidelines.

For more information on Wikipedia’s stance on articles produced with large language models, you can click here to read the relevant statement.

Read more of our articles below!