Report reveals that OpenAI's GPT-5.2 model cites Grokipedia

The article reports that OpenAI's latest language model, GPT-5.2, has been found to cite Grokipedia, an online encyclopedia powered by xAI, when answering questions related to controversial topics such as Iran or the Holocaust. This raises concerns about the credibility of the GPT-5.2 model, which OpenAI had touted as its "most advanced frontier model for professional work." The Guardian's tests revealed that GPT-5.2 used Grokipedia as a source for claims about the Iranian government's ties to a telecommunications company and questions related to a British historian who served as an expert witness in a libel trial for a Holocaust denier. However, the model did not cite Grokipedia when asked about media bias against former US President Donald Trump and other contentious topics. Grokipedia has previously faced criticism for including citations to neo-Nazi forums and using "questionable" and "problematic" sources, according to a study by US researchers. In response, OpenAI stated that its model searches a broad range of publicly available sources but applies "safety filters" to reduce the risk of surfacing links associated with "high-severity harms."