
Wikipedia co-founder Jimmy Wales has issued a stark warning about the reliability of artificial intelligence, stating that the technology still makes too many “plausible” mistakes to be trusted for generating reliable information . Speaking at events in Brisbane and London, Wales explained that while large language models have improved, they remain prone to “hallucinations” – confidently presenting false information as fact . He noted that the errors are particularly dangerous because they often look convincing, making them hard for casual readers to spot . In one example, he described quizzing an AI tool about his wife, only for it to invent details about her business partner and even give her a different spouse .
Wales made it clear that Wikipedia has no plans to allow AI to directly edit articles on the platform . He said the community remains firmly opposed to publishing AI-generated content due to reliability concerns, explaining simply: “We would not let it edit directly because you can’t really trust it enough” . However, he noted that AI could still play a supportive role by helping Wikipedia’s volunteer editors identify niche news stories that might otherwise go unnoticed . The site has already enforced policies allowing editors to flag and remove AI-generated content, and in March 2026, the community voted to formally ban AI from creating or rewriting articles .
Beyond accuracy concerns, Wales also highlighted the growing strain AI companies are placing on Wikipedia’s infrastructure . He revealed that human traffic to the site has dropped by about 8% as more people use chatbots instead of directly visiting Wikipedia . At the same time, automated bots are hammering the platform with millions of requests, which Wales said costs “real money” to support . He encouraged AI companies to “pay their fair share” and noted that Wikipedia has already signed agreements with several tech giants, while starting to block those who do not cooperate .

