British historian Sir Richard Evans, one of the leading scholars of the Third Reich, was astonished to learn a great deal about himself from Grokipedia—the new “AI encyclopedia” launched last week by Elon Musk. According to the platform, Evans supposedly wrote expert witness reports for the libel trial of Holocaust denier David Irving, completed his doctorate under Theodore Zeldin, succeeded David Cannadine as Regius Professor of History at Cambridge (a position established by Henry VIII), and supervised theses on Bismarck’s social policy. As Evans soon discovered, all of this was made up.
Thus began a rocky start for humanity’s latest attempt to compile “the sum of all human knowledge”—or, as Musk put it, to create a compendium of “the truth, the whole truth, and nothing but the truth,” powered by his Grok AI model.
Launching Grokipedia, the billionaire declared it “better than Wikipedia”—or “Wokepedia,” as his supporters call it, implying that the world’s largest online encyclopedia allegedly reflects “left-liberal bias.” One user on X captured the prevailing mood among Musk’s fans: “Elon just buried Wikipedia. Thank God.”
But users quickly noticed that Grokipedia copied entire passages from Wikipedia, contained numerous factual errors, and often reflected Musk’s own political leanings. At the same time, Musk was posting on X about an “inevitable civil war in Britain,” urging Britons to “unite with strong men” like far-right activist Tommy Robinson and claiming that only the AfD could “save Germany.”
Musk became so enthused by his project that he announced plans to engrave the “comprehensive record of knowledge” onto durable oxide and place it “in orbit, on the Moon, and on Mars”—to preserve it for future generations.
Evans, however, found that Musk’s fact-checking AI faced far more down-to-earth problems. “Chat messages there carry the same weight as academic research,” he told The Guardian. According to him, the article on Nazi architect and armaments minister Albert Speer repeated myths long debunked in a 2017 biography, while the entry on Marxist historian Eric Hobsbawm—whose biography Evans himself wrote—claimed that Hobsbawm had lived through Germany’s 1923 hyperinflation and served as a liaison officer, but made no mention of his two marriages.
David Larsson Heidenblad, deputy director of the Centre for the History of Knowledge in Lund, sees in this a clash between two cultures of understanding. “We live in a moment when faith is growing in the idea that algorithmic aggregation is more reliable than human comprehension,” he said. “For Silicon Valley, mistakes are not a failure but part of the process. Academic culture, by contrast, is built on trust, on knowledge tested by time—and on the recognition that no one can know everything.”
Grokipedia continues a centuries-long encyclopedic tradition—from China’s 15th-century Yongle scrolls to the Enlightenment’s Encyclopédie and the Encyclopaedia Britannica to Wikipedia, founded in 2001. But it is the first encyclopedia written largely by artificial intelligence. The central question now is: who controls the truth when the pen is held by machines directed by powerful men?
“If Musk is behind it, I fear political manipulation,” said cultural historian Peter Burke, emeritus professor at Emmanuel College, Cambridge, and author of A Social History of Knowledge. According to him, the anonymity of the articles gives them “a false aura of authority.”
Andrew Dudfield, head of AI at the UK fact-checking organization Full Fact, added: “We have to ask whether we should trust an encyclopedia created by artificial intelligence—a copy of reality filtered through a lens. It demands the same level of trust but offers none of the same transparency. It’s unclear how much human oversight exists, what the AI actually generates, and what data it has been trained on. It’s hard to trust something whose decisions cannot be traced.”