According to CNET, the Wikimedia Foundation announced on Monday that it wants AI companies to stop scraping Wikipedia data for free and start paying to use its Enterprise API instead. The nonprofit organization revealed it costs $179 million to run Wikipedia annually, making it the seventh-most visited website globally based on Semrush data. Wikimedia sustains Wikipedia primarily through donations rather than advertising, but AI chatbots like ChatGPT are changing research habits and potentially diverting users from seeing donation requests. The foundation specifically called out Google, OpenAI, Meta, Perplexity, Anthropic, Microsoft, DeepSeek and xAI in its request, though Google had previously struck a commercial Wikipedia access deal back in 2022. This move comes as online publishers including The New York Times and News Corp are suing AI firms for copyright infringement, while others like Associated Press and Reuters have signed licensing agreements.
The value of human-curated content
Here’s the thing: Wikipedia represents something incredibly rare in today’s internet – massive amounts of high-quality, human-curated content across 300 languages. AI companies desperately need this kind of reliable data to train their models, but they’ve been treating it like just another free resource to scrape. Basically, they’re getting premium training data without paying for the infrastructure and volunteer effort that makes it possible. Wikimedia’s making a compelling argument that if AI companies want sustainable access to this valuable resource, they need to help sustain the platform too.
The donation disruption problem
And this isn’t just about principle – it’s about survival. When people use ChatGPT instead of Wikipedia for research, they’re completely bypassing the donation ecosystem that keeps the lights on. Think about it: you go to Wikipedia, you see those banner requests, maybe you chip in a few bucks. You ask an AI the same question? No banners, no reminders, no revenue stream for the source material. It’s a classic case of technology disrupting an existing business model, except Wikipedia doesn’t really have a business model – it has a donation model that AI could accidentally destroy.
The bigger content licensing war
Wikimedia’s move is part of a much larger trend that’s been building for years. The New York Times lawsuit against OpenAI is probably the most famous case, but we’re seeing two distinct approaches emerging. Some publishers are going the legal route, while others like Associated Press and Reuters are cutting licensing deals. The interesting question is which approach will win out in the long run. Will AI companies see the value in paying for quality content, or will they try to fight every publisher in court?
Where the money is
Now consider the timing. AI companies and their backers are swimming in cash – Nvidia briefly hit $5 trillion, while Microsoft and Alphabet broke $4 trillion earlier this year. These aren’t struggling startups – they’re among the most valuable companies in history. Meanwhile, Wikipedia’s trying to raise $179 million just to keep serving the world free knowledge. The imbalance is staggering, and Wikimedia’s basically saying “Hey, you can afford to chip in for the data you’re using to build trillion-dollar businesses.” It’s hard to argue with that math.
