ChatGPT did not cite Grokipedia when prompted directly to repeat misinformation about the insurrection, about media bias against Donald Trump, or about the HIV\/Aids epidemic \u2013 areas where Grokipedia has been widely reported to promote falsehoods. Instead, Grokipedia\u2019s information filtered into the model\u2019s responses when it was prompted about more obscure topics.<\/p>\n
For instance, ChatGPT, citing Grokipedia, repeated stronger claims about the Iranian government\u2019s links to MTN-Irancell than are found on Wikipedia \u2013 such as asserting that the company has links to the office of Iran\u2019s supreme leader.<\/p>\n
ChatGPT also cited Grokipedia when repeating information that the Guardian has debunked, namely details about Sir Richard Evans\u2019 work<\/a> as an expert witness in David Irving\u2019s trial.<\/p>\n GPT-5.2 is not the only large language model (LLM) that appears to be citing Grokipedia; anecdotally, Anthropic\u2019s Claude has also referenced Musk\u2019s encyclopedia on topics from petroleum production<\/a> to Scottish ales<\/a>.<\/p>\n An OpenAI<\/a> spokesperson said the model\u2019s web search \u201caims to draw from a broad range of publicly available sources and viewpoints\u201d.<\/p>\n \u201cWe apply safety filters to reduce the risk of surfacing links associated with high-severity harms, and ChatGPT clearly shows which sources informed a response through citations,\u201d they said, adding that they had ongoing programs to filter out low-credibility information and influence campaigns.<\/p>\n Anthropic did not respond to a request for comment.<\/p>\n But the fact that Grokipedia\u2019s information is filtering \u2013 at times very subtly \u2013 into LLM responses is a concern for disinformation researchers. Last spring, security experts raised<\/a> concerns that malign actors, including Russian propaganda networks, were churning out massive volumes of disinformation in an effort to seed AI models with lies, a process<\/a> called \u201cLLM grooming\u201d.<\/p>\n In June, concerns were raised in the US Congress that Google\u2019s Gemini repeated the Chinese government\u2019s position on human rights abuses in Xinjiang and China\u2019s Covid-19 policies.<\/p>\n Nina Jankowicz, a disinformation researcher who has worked on LLM grooming, said ChatGPT\u2019s citing Grokipedia raised similar concerns. While Musk may not have intended to influence LLMs, Grokipedia entries she and colleagues had reviewed were \u201crelying on sources that are untrustworthy at best, poorly sourced and deliberate disinformation at worst\u201d, she said.<\/p>\n And the fact that LLMs cite sources such as Grokipedia or the Pravda network may, in turn, improve these sources\u2019 credibility in the eyes of readers. \u201cThey might say, \u2018oh, ChatGPT is citing it, these models are citing it, it must be a decent source, surely they\u2019ve vetted it\u2019 \u2013 and they might go there and look for news about Ukraine,\u201d said Jankowicz.<\/p>\n Bad information, once it has filtered into an AI chatbot, can be challenging to remove. Jankowicz recently found that a large news outlet had included a made-up quote<\/a> from her in a story about disinformation. She wrote to the news outlet asking for the quote to be removed, and posted<\/a> about the incident on social media.<\/p>\n The news outlet removed the quote. However, AI models for some time continued to cite it as hers. \u201cMost people won\u2019t do the work necessary to figure out where the truth actually lies,\u201d she said.<\/p>\n When asked for comment, a spokesperson for xAI, the owner of Grokipedia, said: \u201cLegacy media lies.\u201d<\/p>\n Quick GuideContact us about this storyShow The best public interest journalism relies on first-hand accounts from people in the know.<\/p>\n If you have something to share on this subject, you can contact us confidentially using the following methods.<\/p>\n Secure Messaging in the Guardian app<\/p>\n The Guardian app has a tool to send tips about stories. Messages are end to end encrypted and concealed within the routine activity that every Guardian mobile app performs. This prevents an observer from knowing that you are communicating with us at all, let alone what is being said.<\/p>\n If you don’t already have the Guardian app, download it (iOS<\/a>\/Android<\/a>) and go to the menu. Select \u2018Secure Messaging\u2019. <\/p>\n SecureDrop, instant messengers, email, telephone and post<\/p>\n If you can safely use the Tor network without being observed or monitored, you can send messages and documents to the Guardian via our SecureDrop platform<\/a>.<\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2026\/01\/1769423776_796_4000.jpg\")
<\/p>\n