I named Wikipedia, the free online encyclopedia, back in 2001 and led the project during its first year. The name was obvious: a portmanteau of 'wiki' (Hawaiian for 'quick') and 'pedia' (Greek for 'education').

You can only imagine how cool it was to be a freshly minted philosophy PhD hired to start a free encyclopedia to which everyone could contribute.

In the wake of the 2008 dot-com collapse. I was laid off from the company that ran Wikipedia, along with a handful of other new hires and eventually cut ties with Wikipedia.

Even in those early days, I believed Wikipedia's culture had become dominated by bias and anarchists, which drove out some of its most valuable contributors – the very people a dynamic digital encyclopedia needs most.

But Wikipedia never stopped being important, which is why I later became one of site's most vocal critics.

Eventually, I became tired of critiquing Wikipedia from the sidelines, so I proposed a comprehensive reform plan to the site's leadership called 'Nine Theses on Wikipedia' – a manifesto to reverse its bias.

When no changes emerged, I rejoined Wikipedia as a volunteer to push for reform through a proposed group called WikiProject Intellectual Diversity.

Our goal was to push Wikipedia toward greater intellectual tolerance by promoting genuine neutrality, fair decision-making and transparent governance across Wikipedia's editorial processes.

But in June, I was banned indefinitely from editing entries on Wikipedia for trying to recruit for this group – hardly justifiable in my view, particularly against one of Wikipedia's own founders.

Since being banned, I have often been asked: Where do we go from here? Maybe, I am told, we should just ignore Wikipedia and give up trying to reform its deep-seated bias and intolerance. After all, is Wikipedia even relevant in the age of AI?

It would be a massive error simply to ignore Wikipedia. True, according to the site itself, traffic has been trending downward – from a monthly average of more than one billion visitors between May 2024 and May 2025 to an average of about 815 million visitors since December 2025. 

The most likely explanation for the decline is that AI chatbots such as ChatGPT, Grok and Claude now answer questions directly, reducing the need to visit Wikipedia.

But that is still hundreds of millions of visitors each month – an enormous amount of traffic. By comparison, according to Semrush, CNN.com draws roughly 200 million monthly visits and NYTimes.com roughly 475 million.

Besides, Wikipedia's influence has unquestionably increased since so many AI platforms like Claude and ChatGPT began to train their algorithms on Wikipedia data. Today, you still receive much of the information you ask about the world from Wikipedia, but it also arrives indirectly via the leading AI chatbots now used daily.

Wikipedia's bias is easy to demonstrate. As I see it, the site consistently favors globalist, academic, secular and progressive positions, or as I like to call it, the GASP framework.

This can be seen in the titles that top Wikipedia entries – such as the one called 'Gaza genocide,' which has a clear anti-Israel bent. Or the dismissive 'Cultural Marxism conspiracy theory,' which smears a serious concept discussed by scholars and political commentators as nothing more than a conspiracy.  

An entry detailing President Donald Trump's history of entanglements with Ukraine is titled a 'scandal,' but Biden's history with Ukraine is framed as a theory. The same bias infects any culture-war topic.

And then there are entries around Covid, which are rife with those leading phrases 'theories' and 'conspiracies.' In one crucial entry, we read that the widely-held Covid-19 lab leak theory is 'a highly controversial hypothesis' and, in the second paragraph, 'Many scenarios proposed for a lab leak are characteristic of conspiracy theories.'

Years after the pandemic subsided, the bias remains even as doubt about the leak at the Wuhan Institute of Virology has been almost entirely dismissed.

Because I believe Wikipedia is so one-sided and because it continues to be read directly and used heavily to train AI platforms, defeat is simply not an option. The entirety of AI's future depends on it.

There are four things we can do to help fix Wikipedia, or at least counter the impact of its apparent bias.

First, learn to play Wikipedia's own game against them. If we had just a few hundred people pushing back against Wikipedia's prevailing biases, that would be enough to make Wikipedia genuinely neutral, as it seemingly pretends to be now.

If you're willing, create a Wikipedia account, learn its rules and start editing articles on topics you know well – adding well-sourced alternative viewpoints, correcting one-sided characterizations and participating in discussions where editorial disputes are settled and rules refined.

Second, donate your time and effort to Wikipedia, but stop donating money to the Wikimedia Foundation. This is the nonprofit that runs Wikipedia's servers and raises money on its behalf – over $200 million last year. The Wikimedia Foundation has hundreds of employees and while it does not write the encyclopedia, it administers grants to nonprofits that do. 

Starving it of donations would force the Foundation to refocus its core mission while signaling the seriousness with which the public views its declining neutrality.

Third, speak out about Wikipedia. Complaining publicly is not pointless. If the public comes to the general conviction that Wikipedia has abandoned its neutrality policy, this could impact traffic and affect core AI use policies.

There are still many people who think Wikipedia is perfectly fine and have failed to notice just how frequently one-sided it has become. Tell them that encyclopedias are supposed to let people make up their own minds; all too often, Wikipedia fails this effort.

Share your views on social media, in blogs, among friends and anywhere the message can reach people who still assume Wikipedia is neutral.

Fourth, it is time to press for legislative reform. One of the most serious problems with Wikipedia is its failure to correct libel, failing even to provide article subjects with any effective platform for talking back to Wikipedia editors.

For example, as I have argued last year, Congress may have to create a narrow statutory carve-out exempting the Wikimedia Foundation from powerful Section 230 immunity. Section 230 of the Communications Decency Act currently shields the Foundation from virtually all legal liability for content on the site. 

A narrow carve-out would mean that when someone credibly reports libel and the Foundation fails to act, it could be held legally responsible, just as a traditional publisher would be.

There are many things we can do to combat Wikipedia's institutionalized bias. Remember, this is supposed to be the 'encyclopedia anybody can edit.' That means we all should have a say. If Wikipedia fails to live up to that claim of universality, we can and should push back - hard.

Larry Sanger is the co-founder of Wikipedia. His Nine Theses About Wikipedia can be found at larrysanger.org.