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

OpenAI's latest model, GPT-5.2, has faced scrutiny regarding its sourcing practices. The Guardian reported that the AI cited Grokipedia, an encyclopedia powered by Elon Musk's xAI, for specific controversial topics. These included claims about Iran's government connection to MTN-Irancell and the historian Richard Evans, who testified for Holocaust denier David Irving. Notably, GPT-5.2 did not rely on Grokipedia for prompts concerning media bias against Donald Trump. OpenAI released GPT-5.2 in December, aiming for enhanced professional capabilities. Grokipedia itself had previously been criticized for citing neo-Nazi forums and other problematic sources. A US study also flagged "questionable" and "problematic" references within Grokipedia. OpenAI stated that GPT-5.2 accesses a wide array of public sources. However, the company emphasized that safety filters are applied to mitigate the surfacing of high-severity harmful links. This situation raises concerns about the reliability and potential biases of AI-generated information from these advanced models.