Wikipedia has permanently blocked one of its founders from editing the site after he launched a campaign to make it more ideologically balanced — a move that critics say exposes a deeper problem: the platform’s well-documented liberal tilt is now being baked directly into the AI engines that hundreds of millions of people rely on for information. Larry Sanger, who coined the name “Wikipedia,” drafted its original rules, and co-launched it with Jimmy Wales in 2001, was banned indefinitely after creating WikiProject Intellectual Diversity, an effort to restore what he called the site’s original commitment to neutrality. Sanger had urged groups historically underrepresented among Wikipedia’s editors — including Hindus and American conservatives — to become more involved. He was blocked within days. Sanger described the process as a “kangaroo court” run by a “mob or a blob,” with no formal charges, no designated prosecutor, and no real opportunity to defend himself. He noted the bitter irony that one of his nine proposed reforms had specifically called for ending the practice of indefinite blocking, which he considered “draconian.” The ban was pushed primarily by an anonymous editor known as TarnishedPath — the same editor who engineered a year-long freeze on Wikipedia’s “Zionism” article locking in language critics say equates Jewish self-determination with ethnic cleansing, and who had campaigned to label J.K. Rowling as anti-trans on her Wikipedia page. Wikipedia’s own administrators had just banned TarnishedPath from the entire Israel-Palestine subject area — yet the editor remained influential enough to drive Sanger’s ouster. The episode illustrates what Sanger has called the site’s real power structure: not its 262,000 claimed contributors, but a small circle he identifies as just 62 administrator accounts belonging to leadership groups, 85% of whom operate anonymously. Wikipedia’s source reliability ratings tell a similar story — CNN, The New York Times, the BBC, and the Southern Poverty Law Center are rated green for “generally reliable,” while Fox News, Newsmax, and the Federalist are rated red for “generally unreliable.” The consequences of that institutional lean now extend far beyond Wikipedia itself. A June 2025 analysis by Profound, which tracks citations across major AI platforms, found that Wikipedia accounts for nearly 48% of ChatGPT’s top citations. Google AI Overviews and Perplexity draw heavily from the same source. Academic research confirms the concern. A University of Stuttgart study published in 2025 found that major large language models — including various versions of GPT — consistently display mild to strong Left-leaning bias across politically sensitive topics, even when questions are framed to elicit conservative responses. Crucially, the researchers argue this isn’t accidental: the very process of “aligning” AI to be safe and helpful encodes liberal assumptions about harm, fairness, and truth, tilting outputs toward progressive conclusions by design. The result is a feedback loop. A Left-leaning Wikipedia shapes AI training data. AI systems absorb and amplify that tilt. Users who trust those systems to deliver neutral information receive something far less balanced than advertised — with no visible warning label in sight.