OpenAI’s newest model, GPT-5.2, has been found citing Grokipedia—an AI-generated online encyclopedia launched by Elon Musk’s AI company xAI—in responses to a variety of queries. Independent tests by The Guardian show GPT-5.2 referenced Grokipedia at least nine times across more than a dozen questions, including on topics such as Iranian corporations and the biography of a Holocaust historian. Critics say Grokipedia doesn’t allow traditional human editing and has displayed politically biased or low-credibility content, meaning its inclusion in GPT-5.2 outputs raises misinformation concerns. Experts warn that integrating such sources can subtly influence AI outputs and inadvertently lend legitimacy to questionable content, a risk heightened when models draw from AI-generated rather than human-verified information. OpenAI maintains it uses broad, filtered sources, but the case has sparked wider debate about the transparency and quality of training data in large language models.
In academic partnership with Virscend University, a WASC-accredited institution in Irvine, California.