The AI encyclopedia developed by Elon Musk's xAI, Grokipedia, is increasingly appearing in answers from the competitor OpenAI's ChatGPT.

Grokipedia was initially launched last October. Musk created the platform with the intention of addressing what he perceived as political bias in Wikipedia. However, the platform has been controversial since its launch. Many researchers have pointed out that although Grokipedia has a large amount of content that appears to be directly copied from Wikipedia, it often produces highly subjective or even misleading views on sensitive topics such as social issues, historical explanations, and transgender matters.

This phenomenon of "data leakage" has attracted widespread attention. According to The Guardian, the latest GPT-5.2 model cited entries from Grokipedia nine times when answering various different questions. Notably, ChatGPT seems to have some filtering mechanism: when dealing with high-risk or error-prone topics such as major historical events or epidemiology, it avoids using Grokipedia. However, in some less popular areas, it references information from the platform that has been verified as incorrect by fact-checking organizations.

In response to this phenomenon, an OpenAI spokesperson stated that their model aims to gather information from a wide range of public sources and viewpoints. Since Grokipedia's content has been publicly shared on the internet, it seems inevitable that it will be captured by large models and used as part of their training data. This trend has also raised deep concerns within the industry about AI-generated content "self-reinforcement" and the spread of false information across platforms.