ChatGPT Now Citing Musk's Grokipedia as Source, Raising AI Misinformation Alarm

AI Giants Cross-Pollinating Controversial Content

In an ironic twist of the ongoing rivalry between Elon Musk and Sam Altman, their respective AI products appear to be collaborating in unexpected ways. Tests conducted by The Guardian have revealed that OpenAI's newest GPT-5.2 model is now citing Grokipedia, Musk's AI-generated encyclopedia, as a source when answering questions on sensitive topics.

The investigation found that ChatGPT referenced Grokipedia nine times while responding to questions about Iran's political structure, including claims about telecommunications company MTN-Irancell's government ties. It also cited the platform when providing information about British historian Sir Richard Evans, who served as an expert witness against Holocaust denier David Irving.

Grokipedia's Troubled Foundation

Launched by xAI in October 2025, Grokipedia has faced sustained criticism for its sourcing practices. A Cornell University analysis discovered the platform contains over twelve thousand citations from sources classified as having very low credibility. Researchers found references to the neo-Nazi website Stormfront forty-two times, the conspiracy site Infowars thirty-four times, and the white nationalist outlet VDare over a hundred times.

Unlike Wikipedia's collaborative human editing model, Grokipedia uses a centralised AI system where users can only suggest corrections through a feedback form rather than directly modifying content.

Selective Citation Patterns

Intriguingly, ChatGPT appears to be selective about when it draws from the controversial source. The Guardian noted that the AI avoided citing Grokipedia when asked about the January sixth US Capitol insurrection or media bias against Donald Trump, suggesting the model may filter citation choices based on topic prominence whilst still relying on the source for less scrutinised subjects.

Industry-Wide Concern

The issue extends beyond OpenAI. Anthropic's Claude AI has also been observed using Grokipedia for queries on topics ranging from petroleum production to Scottish ales. Researchers warn that once untrustworthy information enters AI systems, it can spread quietly and become difficult to correct, gaining an undeserved veneer of legitimacy through the authoritative presentation of AI chatbots.