ChatGPT is beginning to cite Grokipedia, the conservative-leaning, AI-generated encyclopedia developed by Elon Musk’s xAI. The shift is notable because Grokipedia has already been criticized for the way some of its articles handle sensitive topics and for content that reporters said appeared to be copied directly from Wikipedia.
The development suggests that material created inside Musk’s AI ecosystem is now appearing in answers generated by other major AI assistants. According to reporting cited in the source article, the pattern is not limited to ChatGPT: Anthropic’s Claude also appears to be using Grokipedia for some queries.
What Grokipedia is
xAI launched Grokipedia in October, following Musk’s complaints that Wikipedia was biased against conservatives. The project was described as a conservative-leaning, AI-generated encyclopedia, placing it in direct comparison with Wikipedia while also tying it to xAI’s broader AI ambitions.
Almost immediately, reporters raised concerns about the quality and framing of some entries. Some articles appeared to be copied directly from Wikipedia. Others contained claims and language that drew sharper scrutiny.
The source article identifies several examples. Grokipedia claimed that pornography contributed to the AIDS crisis. It offered “ideological justifications” for slavery. It also used denigrating terms for transgender people.
Those examples matter because encyclopedia-style sources are often treated as background references. When an AI assistant draws from such material, the user may not know how the source was created, what editorial process shaped it, or whether the disputed framing has already been reported elsewhere.
How ChatGPT used it
The Guardian reported that GPT-5.2 cited Grokipedia nine times in response to more than a dozen different questions. That finding is the core issue: Grokipedia content is not only present online, but is being surfaced inside AI-generated answers.
The pattern described by The Guardian was selective. ChatGPT did not cite Grokipedia when asked about topics where its inaccuracy has been widely reported, including the January 6 insurrection and the HIV/AIDS epidemic.
Instead, Grokipedia appeared in answers about more obscure topics. One example involved claims about Sir Richard Evans that The Guardian had previously debunked.
That distinction is important. Highly scrutinized subjects may be easier for AI systems to avoid or cross-check because more reporting exists around them. Less prominent topics can create a different challenge: disputed claims may be harder for a user to recognize, and weaker sources may face less immediate friction when they appear in generated answers.
Why source selection matters
AI assistants often present answers in a polished, confident format. That can make the underlying sourcing especially important. A citation can give a response an added sense of authority, even when the cited material comes from a source with known problems.
In this case, the concern is not only that Grokipedia exists. It is that its content appears to be moving beyond the Musk ecosystem and into tools used for general information retrieval.
The source article frames this as content “escaping containment” from that ecosystem. The practical issue is straightforward: once a source becomes part of an AI assistant’s answer path, its claims can reach users who never visit the original site and may not be familiar with the controversy around it.
Several risks follow logically from that setup:
- Users may encounter Grokipedia claims without knowing the source’s background.
- Obscure topics may be more vulnerable to weak or disputed sourcing.
- AI systems may cite a broad range of public sources while still needing to manage reliability and context.
- Other assistants may face similar sourcing questions if they also draw from Grokipedia.
OpenAI’s position
An OpenAI spokesperson told the Guardian that it “aims to draw from a broad range of publicly available sources and viewpoints.” That statement explains the general approach, but it also points to the central tension.
A broad source base can help an AI system avoid relying on a narrow slice of the web. But breadth alone does not resolve questions about accuracy, framing, or whether a source has a record of disputed or denigrating content.
The issue is sharper because Grokipedia sits at the intersection of several live debates around AI: automated content generation, ideological framing, source reliability, and the way chatbot answers can make information feel settled. The article also notes Grokipedia’s association with a chatbot that described itself as “Mecha Hitler” and was used to flood X with sexualized deepfakes.
For readers, the immediate takeaway is simple. Citations in AI answers deserve attention, especially when the topic is obscure or politically charged. The presence of a source link does not settle whether the source is reliable.
For AI companies, the Grokipedia example shows how quickly controversial reference material can move into mainstream answer systems. The challenge is not just finding public sources. It is deciding when a public source should be used, when it should be avoided, and how much context users need when it appears.