Skip to content
Submit incident
AII-1001-00867
ReportedReported by fewer than three independent publishers, with no official finding.

ChatGPT's GPT-5.2 model cites Grokipedia as source for misinformation

Tests by the Guardian found that OpenAI's GPT-5.2 model cited Grokipedia, an AI-generated encyclopedia, as a source nine times across queries on topics including Iranian politics and historian Sir Richard Evans, including when repeating information the Guardian had debunked. Grokipedia, launched by xAI, does not allow human editing and has been criticized for propagating right-wing narratives.

24 January 2026 Β· OpenAI Β· Reliability and hallucination
24 January 2026
AII-1001-008672 sources ↓
Quick facts
Category
Reliability and hallucination
Country
United States
Source
The Guardian
Date
24 January 2026
Drafted by
Raidu Incident Detection

What happened

In tests by the Guardian, OpenAI's GPT-5.2 model cited Grokipedia nine times in response to more than a dozen queries on topics including Iranian political structures and the biography of British historian Sir Richard Evans.

Grokipedia is an AI-generated encyclopedia launched in October that does not allow direct human editing, instead an AI model writes content. It has been criticised for propagating right-wing narratives on topics including gay marriage and the 6 January insurrection.

ChatGPT cited Grokipedia when repeating information that the Guardian has debunked, including details about Sir Richard Evans' work as an expert witness in David Irving's trial. Anthropic's Claude also referenced Grokipedia on topics from petroleum production to Scottish ales.

Reported impact

The model cited Grokipedia when asserting stronger claims about the Iranian government's links to MTN-Irancell than are found on Wikipedia, such as asserting that the company has links to the office of Iran's supreme leader.

Index analysis

ChatGPT did not cite Grokipedia when prompted directly to repeat misinformation about the 6 January insurrection, media bias against Donald Trump, or about the HIV/Aids epidemic. Instead, Grokipedia's information filtered into the model's responses when prompted about more obscure topics.

Governance controls

An OpenAI spokesperson said the model's web search 'aims to draw from a broad range of publicly available sources and viewpoints' and that they 'apply safety filters to reduce the risk of surfacing links associated with high-severity harms'.

Classification

Category
Reliability and hallucination
Industry
Media, social platforms and marketing
Country
United States
Harm status
Potential harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Public interest
Harmed
General public
Failure modes
Fabricated citation
Consequences
None recorded
AI systems
GPT-5.2, Claude, Grokipedia

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

The Guardian
Latest ChatGPT model uses Elon Musk's Grokipedia as source, tests reveal
24 January 2026
mint
Is Wikipedia dead? OpenAI's GPT-5.2 turns to Elon Musk's Grokipedia for real-time answers | Mint
25 January 2026