GPT-5.2 cites Musk's Grokpedia, raising misinformation concerns

The latest ChatGPT model, GPT-5.2, has cited Elon Musk’s Grokipedia as a source on various topics, raising concerns about misinformation, The Guardian reported.

Tests showed GPT-5.2 referenced Grokipedia nine times across more than a dozen questions, often on obscure subjects.

Grokipedia is an AI-generated online encyclopedia positioned as Wikipedia rival but does not allow human editing, relying solely on AI content creation.

ChatGPT did not cite Grokipedia when asked about well-documented misinformation areas, such as the US Capitol insurrection or media bias, but did so on less prominent topics.

Experts warn that reliance on sources like Grokipedia could spread false information and lend undue credibility to unverified content.

🔗 Source: The Guardian

🧠 Food for thought

Implications, context, and why it matters.

The full scale of Grokipedia citations remains unquantified

A broader count of how often large language models cite Grokipedia should be run beyond the Guardian’s tests, which found citations nine times across more than a dozen questions.
The share of answers from other models, including Anthropic’s Claude, that reference Grokipedia across topics should be quantified rather than left as anecdote.
OpenAI’s “safety filters” and “ongoing programs to filter out low-credibility information and influence campaigns” should be scrutinized after the Guardian found ChatGPT cited Grokipedia while repeating claims the Guardian says it has debunked about Sir Richard Evans’ work as an expert witness in David Irving’s libel trial.

Concerns about “LLM grooming” could spur demand for verification tools

Tech operators and investors may see a market for products that track and improve large language model (LLM) source integrity.
Companies using AI for business functions could face legal or reputational fallout if models repeat disputed or false claims. The Guardian reports ChatGPT, citing Grokipedia, went beyond Wikipedia on MTN-Irancell and asserted ties between the Iranian government and the office of Iran’s supreme leader.
This risk may raise demand for independent tools that audit the sources an LLM cites, adding a verification layer on top of a platform’s general safety measures.