OpenAI’s newest ChatGPT model, GPT-5.2, has drawn scrutiny after it cited Grokipedia, an AI-generated online encyclopedia linked to Elon Musk’s xAI.
Testing reported by The Guardian found GPT-5.2 referenced Grokipedia on a limited set of prompts, mainly on niche or complex topics. The case comes amid rising competition in the artificial intelligence (AI) market.
The Guardian said its tests identified Grokipedia citations nine times across responses to more than a dozen questions. Researchers noticed the links when prompts moved into technical or less familiar areas. Questions included explanations of Iran’s political system and power centers. Those prompts covered the role of the Basij paramilitary force and influential foundations.
The testing also asked for background on British historian Sir Richard Evans. In those replies, ChatGPT displayed Grokipedia among the sources it cited. What caught attention was the pattern of when the citations appeared. The coverage said the model tended to avoid Grokipedia on high-profile misinformation topics, including Trump-related bias claims and questions about the HIV epidemic.
However, Grokipedia surfaced when prompts involved obscure historical or biographical claims, the report said. In some cases, the chatbot echoed claims aligned with Grokipedia’s framing, and the Guardian said some assertions looked stronger than Wikipedia’s versions.
Grokipedia launched in October 2025 and positions itself as a rival to Wikipedia, the reports said. The platform also markets itself as a leading alternative to Wikipedia’s model. Unlike Wikipedia, Grokipedia uses automated systems to produce and refresh its entries. Users can request topics, but they do not edit pages in the way Wikipedia contributors do.
Some supporters say this approach can scale faster and cover more subjects. Critics, however, warn that minimal human oversight may increase the risk of errors and perceived political bias.
Since launch, Grokipedia has faced criticism for allegedly promoting right-leaning narratives on contested issues. The cited examples included same-sex marriage and the January 6 United States (US) Capitol attack.
The Guardian’s coverage also described Grokipedia as part of xAI’s effort to build an alternative to Wikipedia. It noted claims that Wikipedia has become “less centrist” in recent years.
Recent reports cited in that coverage said Grokipedia has created over 6 million articles. The sources said that total exceeds 80% of English Wikipedia’s article count.
OpenAI said ChatGPT draws from a broad mix of publicly available sources and viewpoints. A spokesperson told The Guardian that OpenAI applies safety filters to reduce links tied to “high-severity harms.”
In addition, the spokesperson said ChatGPT aims to show which sources informed an answer through citations. OpenAI said it uses safety measures to limit harmful or misleading content while aiming to attribute information clearly.
Meanwhile, the Guardian reported similar Grokipedia references in Anthropic’s Claude chatbot. Those citations appeared in answers ranging from global oil production to regional food and drink, including Scottish ales.
Moreover, the debate adds pressure on AI firms as AI-generated encyclopedias expand online. Furthermore, the reports pointed to growing regulatory and public demands for transparency and accountability.
Still, the selective appearance of Grokipedia in niche answers may keep attention on GPT-5.2 Grokipedia citations. OpenAI’s statement suggests it will keep using diverse public sources while applying safety filters and visible attribution. Consequently, the debate may keep regulators focused on citation controls.