Elon Musk’s Grokipedia Inspires More Than Just ChatGPT in the Chatbot Arena

ChatGPT is tapping into Grokipedia, and it’s not the only AI tool doing so. Citations from Elon Musk’s AI-driven encyclopedia are popping up in responses from various AI systems, raising concerns about accuracy and misinformation as Musk aims to reshape perceptions.

Since its launch last October, Grokipedia has remained a relatively minor information source. According to Glen Allsopp, a marketing strategist at an SEO firm, testing revealed Grokipedia was referenced in over 263,000 ChatGPT responses from 13.6 million prompts, citing around 95,000 individual pages. In contrast, English-language Wikipedia appeared in 2.9 million responses. Allsopp noted that while Grokipedia is still far behind, its performance is impressive for a newcomer.

Data from a marketing platform tracking billions of citations indicates Grokipedia receives about 0.01 to 0.02 percent of all ChatGPT citations daily—a small but growing share since mid-November. Another analytics tool found a similar increase in Grokipedia’s visibility in AI responses from December, though it remains a secondary source compared to established platforms like Wikipedia.

Grokipedia citations are more prevalent in ChatGPT than in other AI tools being monitored. However, a similar uptick was noted in Google’s AI products in December. Allsopp reported Grokipedia was referenced in approximately 8,600 Gemini answers, 567 AI Overviews, 7,700 Copilot responses, and 2 Perplexity answers, with significant drops in Gemini and Perplexity from the previous month. While no firms are tracking citations for Anthropic’s Claude, anecdotal evidence suggests it also references Grokipedia.

AI tools often cite Grokipedia for niche or specific factual queries. Analysts, including Jim Yu from an analytics firm, noted that ChatGPT and AI Overviews typically use Grokipedia for “non-sensitive queries” like definitions, although they differ in how much authority they grant it. Yu observed that in AI Overviews, Grokipedia usually appears alongside other sources as a supplementary reference, while ChatGPT tends to treat it as a primary source.

Experts caution that using Grokipedia can spread misinformation and promote biased viewpoints. Unlike Wikipedia, which is edited by humans through a transparent process, Grokipedia is generated by xAI’s chatbot Grok, known for controversial statements and problematic content. Initially, many of Grokipedia’s articles were direct copies of Wikipedia, but others reflected harmful views. For instance, articles about Musk downplay his family’s wealth and controversial history, while the entry on “gay pornography” inaccurately links it to the HIV/AIDS epidemic. The article on US slavery includes problematic ideological justifications, and editing is overseen by Grok, which is also flawed. Grokipedia is particularly vulnerable to “LLM grooming,” or data poisoning.

An OpenAI spokesperson stated that ChatGPT aims to draw from a wide range of publicly available sources and allows users to assess the reliability of those sources. They emphasized the application of safety filters to minimize harmful content. A spokesperson from Perplexity refrained from commenting on the risks of citing AI-generated material but highlighted their focus on accuracy. Anthropic and xAI did not provide comments, and Google declined to respond.

Ultimately, Grokipedia cannot be reliably cited as a source, regardless of its frequency in AI responses. It operates as an AI-generated system without human oversight, often relying on unverifiable material. This raises the risk of perpetuating biases and errors, as noted by experts who warn that fluency can be mistaken for reliability.

Grokipedia may seem credible within its own context, but experts express skepticism about its acceptance as a serious reference by major platforms like Google or OpenAI.