Elon Musk launched Grokipedia last October as an anti-woke, AI-powered Wikipedia. It is, Renée DiResta wrote for The Atlantic at the time, part of “an escalating campaign to discredit Wikipedia and reshape what counts as a reliable source of basic information in the age of AI.”
Grokipedia allows humans to suggest edits, but Columbia’s Tow Center for Digital Journalism found in February that the tool was increasingly editing itself. “Grok-supplied edits spiked in December and have overtaken human submitters, making up more than three-quarters of the suggestions,” Tow Center fellow C.J. Robinson wrote.
Now, though, it appears that Grokipedia isn’t being edited at all. “The encyclopedia stopped editing itself and processing suggested edits from humans,” DiResta and the Stanford Cyber Policy Center Ronald Robertson wrote for Lawfare this week. “As far as we can tell, no entry has changed in more than three months.”
That matters, they say, because people actually use Grokipedia, and AI systems, especially ChatGPT, rely on its content:
The significance here extends beyond one internet encyclopedia. Grokipedia is not just a small xAI side project. The web traffic analytics firm Similarweb estimates that Grokipedia received 6.7 million visits in June 2026 and ranked 11,022nd among websites globally — substantial reach for a site launched less than a year earlier. Individual high-profile entries command significant readership: Grokipedia’s page-statistics API — which serves per-article view counts that are not displayed on the front-end website — records over 83 million views for President Obama and 16 million lifetime views for its Musk entry since October 2025.
Grokipedia’s contents also do not remain confined to Grokipedia. An Ahrefs analysis published in March 2026 found the site surfacing in roughly 356,000 citations across AI systems — most often in ChatGPT and Google’s AI Mode, and at lower volumes in Gemini, Copilot, and AI Overviews. That remains a small fraction of Wikipedia’s roughly 25 million such citations, but it is not trivial.
A reference work whose contents are silently frozen and whose editorial record is inaccurate could propagate incorrect or outdated information into answers that people encounter elsewhere. There are also potential reputational consequences for individuals with unreliable Grokipedia biographies who can no longer appeal to the AI for an edit…
As agentic AI systems begin to generate and maintain reference knowledge independently, the question is not only whether their output is accurate at a given moment. It’s whether their behavior over time is observable, and their record stable enough to be accountable for their contributions to the public information ecosystem. At the moment, Grokipedia is neither.
Boing Boing notes that Wikipedia is updated about a quarter of a million times per day.
