ChatGPT has started quoting Musk's Grokipedia

Kira Aronova
Kira Aronova Journalist
ChatGPT has started quoting Musk's Grokipedia
Elon Musk's alternative to Wikipedia - Grokipedia. Photo Getty Images
The new version of ChatGPT uses Grokipedia as a source, sparking debate about the reliability of AI encyclopaedias.

The latest ChatGPT model has started using Grokipedia, an AI-powered online encyclopaedia created by Elon Musk's company xAI. This was reported by The Guardian based on the results of its own tests.

During testing, the GPT-5.2 version repeatedly referred to Grokipedia in its answers to questions about Iran's political and economic systems, as well as in materials related to Holocaust denial. In some cases, the wording from Grokipedia was harsher or less verified than the corresponding information on Wikipedia.

At the same time, ChatGPT did not use direct references to Grokipedia on topics where errors in this AI encyclopaedia had been previously identified. Despite this, its data was still integrated into the answers — primarily on highly specialised or little-known issues that rarely come into the focus of public discussion.

OpenAI explained that the ChatGPT web search system works on the basis of a wide range of open sources and applies security filters to minimise the risk of spreading false or harmful information. The company also emphasised that users are shown the sources used in forming the answers.

Disinformation experts warn that when artificial intelligence quotes Grokipedia, users may automatically perceive it as a reliable source, even if the published data is questionable. In response, xAI stated that it is traditional media that often disseminates inaccurate or distorted information.

What is known about Grokipedia:
Grokipedia is an online encyclopaedia whose content is entirely generated by artificial intelligence. Users can submit requests for corrections, but they cannot edit articles themselves, as they can on Wikipedia.

Elon Musk announced the creation of his own AI encyclopaedia on 6 October, presenting Grokipedia as a "massive improvement on Wikipedia." The idea is that the platform analyses open sources, identifies errors, lack of context or partial truths, and automatically corrects the material.

Grokipedia was officially launched on 27 October. Currently, the encyclopaedia has over 885,000 articles. At the same time, a number of media outlets — including Business Insider, NBC News, Wired and The Verge — have already pointed out factual errors, inaccurate statements and cases of plagiarism from Wikipedia in its materials.

ChatGPT, Grokipedia, Elon Musk, artificial intelligence, misinformation

Share tittle
Science
In Japan, rare crested ibises, which had become extinct in the country several decades ago, have been reintroduced into the wild
Science

In Japan, rare crested ibises, which had become extinct in the country several decades ago, have been reintroduced into the wild

For the first time in several decades, eight crested ibises – a species that became extinct in Japan in the last century – have been released into the wild.

01.06.2026
A nature reserve is to be established near the Ukrainian Antarctic research station 'Akademik Vernadsky'
Science

A nature reserve is to be established near the Ukrainian Antarctic research station 'Akademik Vernadsky'

Ukraine has secured the establishment of a new specially protected area in Collins Bay, located near the Akademik Vernadsky station in Antarctica. The initiative was approved following three years of intensive international negotiations.

25.05.2026
The State Special Communications Service is calling for increased funding for Ukrainian crypto technologies
Science

The State Special Communications Service is calling for increased funding for Ukrainian crypto technologies

The State Service for Special Communications and Information Protection of Ukraine plans to develop a domestic industrial base for cryptographic solutions and to support research institutions.

25.05.2026
Artificial intelligence will be able to create its own next generations by the end of 2028
Science

Artificial intelligence will be able to create its own next generations by the end of 2028

Over the next 12 months, artificial intelligence, working in collaboration with humans, could make a scientific discovery that wins a Nobel Prize.

22.05.2026
A laboratory has opened in Japan where medical experiments are carried out by robots
Science

A laboratory has opened in Japan where medical experiments are carried out by robots

Tokyo University of Science in Japan has opened a laboratory where medical experiments are carried out by robots without human involvement.

11.05.2026