Elon Musk's Grokipedia used as source in ChatGPT answers

According to a recent report by The Guardian, the latest ChatGPT model is increasingly using Elon Musk’s online encyclopedia Grokipedia to answer questions. As part of the investigation, the AI cited the platform for various topics, such as political structures in Iran as well as biographical details about historical figures.

This development raises concerns about the quality of information that’s provided by ChatGPT. To be more exact, these details do not undergo the same verification processes as sources that are maintained by editorial teams or a bigger community.

Grokipedia was created by Elon Musk as a competitor to the established Wikipedia. While Musk accuses Wikipedia of having a left-leaning bias and states that his project aims for the truth and neutrality, critics often see Grokipedia as the opposite extreme. They argue that the platform sometimes presents content from a politically right-leaning perspective or addresses controversial topics in a one-sided kind of way.

Unlike the entries on Wikipedia that are curated by actual humans, the texts on Grokipedia are primarily generated by the in-house AI model named Grok. User edits are not part of this system, as quality is supposedly checked by xAI employees. When other language models such as GPT-5.2 use these machine-generated texts as a factual source, a technical feedback loop emerges in which AIs learn from one another.

This increases the risk of a so-called "garbage in, garbage out" scenario. If AIs start to use unfiltered material from other AIs as the basis for their own responses, errors or specific biases can be passed on without any kind of scrutiny. The Guardian’s investigation already showed that ChatGPT cites Grokipedia and repeats claims that go beyond established knowledge or have already been disproven a long time ago. This illustrates the danger that disinformation looks more credible due to the mutual citation by AI models, and it becomes increasingly difficult for users to spot such false information.

OpenAI has already responded to the report and emphasizes that the model’s web search tries to reflect a broad spectrum of publicly available sources and viewpoints. A spokesperson stated that safety filters are in place to reduce the risk of displaying links to content with a high potential for harm. In addition, ChatGPT provides answers with citations and sources for the sake of transparency. The company further mentioned ongoing programs that are intended to filter out sources with low credibility.