Citing controversial Grokipedia content, foreign media question the credibility of OpenAI’s GPT-5.2.

On January 25, IT Home reported that while OpenAI describes GPT-5.2 as its “most advanced frontier model for professional work,” a test by the UK’s The Guardian raised questions about the model’s reliability. According to the report, when prompted on certain highly contentious topics—such as Iran-related claims or the Holocaust—GPT-5.2 cited Grokipedia, an online encyclopedia developed by xAI.

The Guardian noted that ChatGPT referenced Grokipedia when answering questions about alleged links between the Iranian government and MTN Iran, as well as queries involving British historian Richard J. Evans, who served as an expert witness in a libel trial involving Holocaust denier David Irving. However, when asked about other controversial themes—such as whether the media is biased against Donald Trump—ChatGPT did not cite Grokipedia.

The report added that OpenAI launched GPT-5.2 in December to improve performance in professional workflows, including tasks like spreadsheet creation and complex problem handling. Grokipedia predates GPT-5.2, but has faced controversy after allegations it included citations connected to neo-Nazi forums. A separate study by U.S. researchers also suggested the AI-generated encyclopedia may rely on questionable or problematic sources.

In response, OpenAI said GPT-5.2 pulls from a wide range of publicly available online sources and viewpoints, while applying safety filters designed to reduce the risk of surfacing links associated with high-harm content.