Report reveals that OpenAI’s GPT-5.2 mannequin cites Grokipedia

OpenAI could have known as GPT-5.2 its “most superior frontier mannequin for skilled work,” however checks performed by the Guardian forged doubt on its credibility. In response to the report, OpenAI’s GPT-5.2 mannequin cited Grokipedia, the web encyclopedia powered by xAI, when it got here to particular, however controversial subjects associated to Iran or the Holocaust.

As seen within the Guardian‘s report, ChatGPT used Grokipedia as a supply for claims concerning the Iranian authorities being tied to telecommunications firm MTN-Irancell and questions associated to Richard Evans, a British historian who served as an knowledgeable witness throughout a libel trial for Holocaust denier David Irving. Nevertheless, the Guardian famous ChatGPT did not use Grokipedia when it got here to a immediate asking about media bias towards Donald Trump and different controversial subjects.

OpenAI launched the GPT-5.2 model in December to higher carry out at skilled use, like creating spreadsheets or dealing with complicated duties. Grokipedia preceded GPT-5.2’s launch, however bumped into some controversy when it was seen including citations to neo-Nazi boards. A examine accomplished by US researchers additionally confirmed that the AI-generated encyclopedia cited “questionable” and “problematic” sources.

In response to the Guardian report, OpenAI informed the outlet that its GPT-5.2 mannequin searches the online for a “broad vary of publicly accessible sources and viewpoints,” however applies “security filters to scale back the danger of surfacing hyperlinks related to high-severity harms.”