For years, artificial intelligence tools have been quietly learning from Wikipedia, pulling facts, dates, and explanations to power chatbots and search features.
Now, as the free online encyclopedia turns 25, it’s drawing a line and saying: if you’re going to use our knowledge at scale, it’s time to help pay for it. This week, the Wikimedia Foundation, which runs Wikipedia, announced new partnerships with major AI companies including Amazon, Meta, Microsoft, Perplexity and France-based Mistral AI. These companies will now pay to access Wikipedia’s content in ways that suit large-scale AI training. Why AI depends so heavily on Wikipedia Google had already signed a similar agreement back in 2022, while smaller firms like Ecosia joined later. Together, these deals mark a big shift in how Wikipedia earns money from the tech industry that depends heavily on its data.
Wikipedia is one of the richest collections of human-written knowledge on the internet, with more than 65 million articles available in over 300 languages. Because the content is carefully edited by volunteers and adheres to strict sourcing rules, it is particularly useful for training AI systems that require reliable and neutral information. That has made Wikipedia a key ingredient in how today’s chatbots and AI assistants learn about the world, from history and science to pop culture and politics. Also read: Viral ‘Are You Dead?’ app now worth ₹13 crore, highlights rising loneliness in China
When free starts becoming expensive While anyone can read Wikipedia for free, running the website is far from cheap. The foundation says that aggressive data scraping by AI bots has increased server traffic and infrastructure costs, putting extra pressure on a platform that mostly relies on small public donations. At the same time, human visitors are slowly dropping off. In 2024, Wikipedia reported an 8% fall in pageviews, partly because people now get answers directly from AI tools instead of clicking through to websites. So while AI companies were benefiting from Wikipedia’s content, Wikipedia itself was left dealing with higher costs and fewer readers. “Happy to help — but pay your share” Wikipedia founder Jimmy Wales says he is not against AI companies using the site’s content for training. In fact, he welcomes it, as long as it is done fairly. I’m very happy personally that AI models are training on Wikipedia data because it’s human-curated. But he also made it clear that the burden should not fall on ordinary donors. You should probably chip in and pay for your fair share of the cost that you’re putting on us.
He even joked that he would not trust an AI trained only on social media posts, saying that could result in “a very angry AI”. Also read: NASA astronauts back on Earth safely after unprecedented medical emergency
What exactly are companies paying for? Wikipedia isn’t selling exclusive content. Instead, it is offering Enterprise access, which means: So AI companies still get the knowledge, just in a more organised, paid way. How will users and editors benefit? The Wikimedia Foundation says the money will help improve the platform and tools for volunteers. Some plans include: Jimmy Wales explained the future idea like this: You could ask a question in the Wikipedia search box and it would quote the exact paragraph that answers it. That means AI may actually improve how people use Wikipedia, instead of replacing it.
Also read: Delete these screen-sharing apps from your phone, govt issues warning
Not worried about AI-written Encyclopedias Wales also dismissed concerns about AI-generated alternatives like Elon Musk-backed “Grokipedia”, saying large language models are still not good at writing proper reference material. “Large language models aren’t good enough to write really quality reference material… the more obscure the topic, the worse it gets.” He added that most such projects end up repeating Wikipedia anyway. Why this deal matters for the internet Wikipedia is built on free, volunteer-made knowledge, but AI companies now use it at a massive commercial scale. These new deals make big tech share the cost of running the platform.
Wikipedia stays free for readers, but training AI on public knowledge will no longer be free. If AI companies make billions using free internet knowledge, who pays to keep that knowledge alive? Now, at least some of that cost is shifting to the companies that benefit the most.
The post appeared first on .
