Open AI's GPT 5.2 model has started setting Grokapedia, Elon Musk's AI generated encyclopedia, as a source of its responses. Now. The Guardian ran tests last week and found ChatGPT reference Grokapedia 9 times across more than a dozen user queries, including questions about Iranian political structures and a British historian who helped convict A Holocaust denier. The citations appeared selectively, surfacing an obscure topics while avoiding well documented misinformation
subjects. Grocopedia launched just three months ago with 0 human. Editors and critics have already flagged it for promoting debunked claims, setting neo Nazi forums and whitewashing extremism. Now, how did an AI encyclopedia built by one of Open AI's biggest rivals end up feeding answers back into Open AI's most popular chatbot? Agracopedia now holds over 5.6 million articles, roughly 80 the size of English Wikipedia, and it accomplished that scale in a fraction of the time Wikipedia
needed. Now today we're going to cover what the Guardians testing revealed with specific false information chatt GPD pulled from Gakapedia, why other AI assistants like Claude Anthropics Bot are also setting the platform, and what this means for the integrity of AI generated knowledge. Open AI has responded with a statement about safety filters and source diversity, while XAI simply replied with three words. I'll tell you about those later, right after this short break now.
Open AI marketed GPD 5.2 as its most advanced model for professional work when it launched in December, But the Guardian's investigation raises serious questions about where the professional grade information actually comes from. The tests revealed a clear pattern and how ChatGPT selects Grokapedia as a source, and the implications cut deeper than a
simple citation problem. Users asking complex questions about geopolitics and history are receiving answers built on an AI generated foundation that has already been documented
spreading false claims. Now the guardians were testing focused on a range of topics to see when ChatGPT would pull from Grokopedia versus other sources, and the chatbot did not say Grokopedia. When asked about widely known misinformation, such as false claims about media bias against Donald Trump or debunk theories about HIV and AIDS, ChatGPT avoided the obvious traps. The system instead cited Gracopedia on obscure geopolitical and historical questions. The kind of query is where users
are less likely to catch errors. Specific topics included salaries of Iran's paramilitary force, ownership of the mastas, a fan fountain, and connections between the Iranian government and telecom company MTN Iran Cell. Now, Chachibti made bolder claims about those Iranian government links than Wikipedia does.
Those claims came directly from Grokopedia, and the most troubling example involved Sir Richard Evans, a Cambridge historian who served as the key expert witness in the libel trial wrought by Holocaust denier David Irving. Evans personally examined his own Grokopedia entry and reported that it contained numerous false statements about his career and background. He told the Guardian that all the facts in his entry were wrong. The already runs pretty deep here.
Evans spent years meticulously documenting how Irving falsified evidence about the Holocaust and manipulated historical records. The High Court judge ultimately branded Irving a racist, an anti semite and falsifier of history based largely on Evans expert testimony. ChatGPT cited that same Gracopedia page when responding to user questions about Evans effectively laundering debunked information through Open AI's
flagship product. The historian who exposed the Holocaust deniers lies is now having lies spread about him through AI systems that treat machine generated content as the equivalent of peer reviewed scholarship. Now I've been going through all of our stats here at the show and 39% of you are following the channel and I'm so grateful for you. 61% of you have listened to the show before but haven't hit
the subscribe or follow button. If that's you, if you listen to the show and you're into it, please take a second of your time. You can hit the follow button. I'm going to spend the next 10 years doing this. So that's what you get from me, better content over the next 10 years. All I need is a second of your time. So thank you so much for doing that. Agrocopedia launched on October 27th, 2025 as Elon Musk's answer to what he calls Woke Pedia.
The platform uses XA is Grok large language model to generate articles rather than relying on human editors. Users can suggest corrections through a pop up form, but they can't directly edit content the way Wikipedia allows you to. Musk position the project as a bias free alternative that would purge propaganda from public knowledge. The reality has looked pretty different though. Analysis found Wikipedia articles aligned with Musk's personal views on topics like gender identity.
Tesla Controversies and Neuralink and the Guardian doing their research here, and multiple other outlets documented pages that promote white nationalist talking points, praise neo Nazis, cast Holocaust deniers in positive terms, and frame far right figures as victims of institutional suppression. The scale of Rockopedia's growth
presents its own problems too. Platform reached 5.6 million articles by early 2026, jumping from 800,000 at launch to nearly 80% the volume of English Wikipedia in about 3 months. Human editors and Wikipedia spent over 2 decades building their corpus through careful research, citation and community debate. Grakopedia achieved similar volume by having an AI system scrape existing content, rewrite those articles, and generate new ones at machine speed.
Symmetry is copied Wikipedia near the word for word under a Creative Commons license. While others introduced entirely new claims from questionable sources, researchers from Grokopedia cited neo Nazi forums and its reference lists. Problem that becomes exponentially worse when other AI systems start treating Grokopedia as a legit source.
Now, Jet GBT was not the only AI assistant caught citing Grokopedia. The Guardians testing found the Anthropics. Claude, my favorite AI right now, also referenced the platform when using questions about global oil production and Scottish beer. Little bit different, though. Anthropic did not respond to requests for comment. Because that's what they usually do. Big corporations do not respond because they just wanted to blow over.
And the fact that multiple leading AI systems are pulling from an encyclopedia built by a competitor and built by an AI raises questions about how these comanies curate and filter their web search sources. Open AI told the Guardian that Chat GT aims to draw from a broad range of ublicly available sources and viewpoints, adding that safety filters reduce the risk of servicing links associated with high severity harms.