Photo: Screenshot from the Grokpedia page
Elon Musk’s new AI-generated encyclopedia, Grokipedia, has come under fire just days after its launch, as academics discovered major factual inaccuracies, political bias and a lack of scholarly standards, The Guardian reports. Critics say the platform treats comments from chat rooms and blogs on par with verified academic research.
Sir Richard Evans, a leading British historian, checked his profile on Grokipedia only to find that most of the information about him was false — including invented titles, posts and academic affiliations. “Chat-room contributions are given the same status as serious academic work. The AI just hoovers everything up,” Evans told The Guardian.
Musk promoted Grokipedia as “a repository of the whole truth and nothing but the truth,” claiming it would surpass Wikipedia, which some of his supporters mock as “Wokepedia.” He even said he aims to “engrave the complete collection of knowledge in stable oxide” and store copies on orbit, the Moon and Mars.
However, users quickly spotted that more than 885,000 articles contain extensive copy-and-paste material from Wikipedia — including pages about the PlayStation 5, Ford Focus and Led Zeppelin — while others distort facts or push narratives aligned with Musk’s own views.
The page on Russia’s invasion of Ukraine uses Kremlin-style language about “denazification” and “protecting ethnic Russians,” despite Wikipedia clearly labelling such claims baseless and imperialist. Another entry describes the far-right group Britain First as a “patriotic political party,” ignoring its leader’s jail sentence for hate crimes.
Grokipedia also refers to the January 6 Capitol attack as a “riot,” omitting references to an attempted coup attempt, and gives credence to the racist “white replacement” conspiracy theory. In coverage of Donald Trump’s falsified-documents case, it notes the trial occurred in a “Democrat-majority jurisdiction” while skipping key factual context.
Peter Burke, a Cambridge cultural historian, warned of the risks: “If Musk is behind this, I fear political manipulation. Not all readers will notice distortions, and anonymous entries create a false sense of authority.”
David Larsson Heidenblad of Lund University added that Silicon Valley thinking clashes with academic rigor: “Its culture treats mistakes as a feature, not a flaw. Academia is about building trust over time.”
Andrew Dudfield of fact-checking organization Full Fact highlighted the system’s opacity: “It’s unclear how much is AI-generated, how much is human-written, and what data it’s trained on. Trust requires transparency.”
While Musk once supported Wikipedia, he has repeatedly attacked it since 2023, calling it “biased” and joking he’d pay £1 billion for it to rename itself “Dickipedia.”
Musk’s xAI did not respond to The Guardian’s request for comment. The Wikimedia Foundation stated that Wikipedia’s strength lies in “transparent policies, strong volunteer oversight and a culture of improvement.”
“It exists to inform billions — not to push opinions,” a spokesperson said.
