From ChatGPT to Google, AI tools are starting to cite Grokipedia

Grokipedia

• Grokipedia is increasingly cited by AI tools like ChatGPT and Gemini.
• Researchers worry about Grokipedia's accuracy and lack of human oversight.
• Grokipedia's share of citations is small but steadily rising since its launch.

Did our AI summary help?

Grokipedia, the AI-generated encyclopedia created by Elon Musk’s xAI, is beginning to surface as a cited source across major AI tools — including ChatGPT, Google’s Gemini, AI Overviews, and AI Mode — prompting growing concern among researchers about accuracy and misinformation. A report by The Verge detailed how AI tools are tapping Grokipedia.

Launched in late October, Grokipedia remains a relatively minor reference compared to Wikipedia. According to the report by The Verge, multiple analytics firms say its presence in AI-generated answers is increasing. According to Ahrefs, Grokipedia has appeared in more than 263,000 ChatGPT responses from a dataset of 13.6 million prompts, drawing on roughly 95,000 individual pages. By contrast, English-language Wikipedia was cited in around 2.9 million responses.

“They’re quite a way off, but it’s still impressive for how new they are,” said Glen Allsopp, head of marketing strategy and research at Ahrefs, told The Verge.

Data from marketing platform Profound shows Grokipedia accounts for around 0.01 to 0.02 percent of ChatGPT citations per day — a small share, but one that has steadily climbed since mid-November. Semrush has observed a similar uptick across Google’s AI products since December, though Grokipedia still trails far behind established reference sites, as per the report.

ChatGPT appears to cite Grokipedia more frequently than other platforms. Ahrefs’ analysis found Grokipedia references in around 8,600 Gemini answers, 567 AI Overviews responses, 7,700 Microsoft Copilot answers, and just two Perplexity answers. Several researchers also say anecdotal evidence suggests Anthropic’s Claude may be citing Grokipedia, though this is not formally tracked, as per the report.

In many cases, AI tools turn to Grokipedia when answering niche or highly specific factual questions. Jim Yu, CEO of analytics firm BrightEdge, told The Verge that Grokipedia is typically used for “non-sensitive queries” such as definitions or encyclopaedic lookups. However, how much authority platforms grant it varies.

For Google’s AI Overviews, Grokipedia usually appears alongside multiple sources and is treated as supplementary. ChatGPT, by contrast, often gives Grokipedia greater prominence, sometimes listing it among the first sources cited.

That difference matters, experts say, because Grokipedia lacks the safeguards that underpin Wikipedia. The site is generated and edited by xAI’s chatbot Grok, without transparent human oversight. When it launched, many articles were cloned from Wikipedia, while others reflected racist, transphobic, or misleading views. Critics have also pointed to entries that downplay Musk’s family wealth, mischaracterise historical events, or include questionable ideological framing, as per the report.

Because Grokipedia is AI-generated, it is also vulnerable to “LLM grooming” or data poisoning, where flawed or biased material is repeatedly reinforced through machine-generated citations.

OpenAI said ChatGPT aims to draw from a wide range of publicly available sources and applies safety filters to reduce high-risk content. “ChatGPT clearly shows which sources informed a response through citations, allowing users to explore and assess the reliability of sources directly,” an OpenAI spokesperson told The Verge.

Google declined to comment. xAI did not respond to a request for comment. Perplexity said its focus remains on accuracy, while Anthropic declined to comment on the record, as per the report.