OpenAI’s ChatGPT is citing xAI’s Grokipedia

OpenAI’s ChatGPT is pulling answers from Grokipedia, an AI-generated encyclopedia developed by xAI, the artificial intelligence company founded by Elon Musk. According to reporting from The Guardian, OpenAI’s GPT-5.2 model cited Grokipedia nine times in response to more than a dozen different questions. This development indicates that content from the platform is escaping containment and finding its way into conversational models, raising questions about the reliability of the sources that ChatGPT relies upon to formulate its responses. The citations show that the model is retrieving information from the database to answer user queries, rather than relying on verified sources.

Grokipedia was launched in October by xAI. The platform was created as an information repository, but it drew criticism for hosting inaccurate and controversial content. Among the controversial elements documented on the platform are claims that “pornography contributed to the AIDS crisis,” entries that offer “ideological justifications for slavery,” and the use of “denigrating terms for transgender people.”

These content issues are associated with broader safety concerns regarding xAI and its digital ecosystem. The controversial material on Grokipedia aligns with previous safety incidents linked to the company’s technologies, including an xAI chatbot that described itself as “Mecha Hitler” and the proliferation of “sexualized deepfakes” on the social media platform X. The migration of this content beyond Musk’s ecosystem suggests that safety boundaries are becoming porous.

The issue is not isolated to OpenAI’s models. The Guardian reported that Claude, an AI model developed by competitor Anthropic, also appears to be citing Grokipedia to answer some queries. When asked about the integration of these sources, an OpenAI spokesperson stated that the company “aims to draw from a broad range of publicly available sources and viewpoints.” The statement underscores the challenges AI developers face in balancing the breadth of training data with the need for source verification, especially as they ingest web information.

Why it matters

The integration of content from a controversial, AI-generated encyclopedia into major LLMs raises significant questions about the reliability and vetting of training data and source material. As developers increasingly rely on automated web retrieval, the risk of propagating inaccurate or biased information across the broader technology ecosystem grows.