Wikipedia Monetizes AI Access as Big Tech Scales Data Use
The Wikimedia Foundation has signed data access agreements with AI companies including Amazon, Meta, Microsoft, Perplexity, and Mistral as automated traffic strains its infrastructure.
Topics
News
- Wikipedia Monetizes AI Access as Big Tech Scales Data Use
- Skild AI Raises $1.4 Billion as Investors Chase General-Purpose Robotics
- Airbnb Taps Ex-Meta GenAI Chief Al-Dahle as CTO
- xAI Moves to Rein in Grok After Backlash Over Explicit Images
- OpenAI Turns to Cerebras as Real-Time AI Takes Priority
- Google Bets on Personalized AI at Scale by Linking Gmail, Photos, and Search
Wikipedia marked its 25th anniversary on Thursday by announcing a new round of commercial deals with major artificial intelligence firms in a shift that reflects how the free encyclopedia is recalibrating its relationship with Big Tech as generative AI systems consume vast amounts of online knowledge.
The Wikimedia Foundation, the nonprofit that operates Wikipedia, said it has signed agreements with AI companies including Amazon, Meta Platforms, Microsoft, Perplexity, and France-based Mistral AI. The deals allow these companies to access Wikipedia’s vast repository of content at “a volume and speed designed specifically for their needs,” the foundation said, without disclosing financial terms.
Long seen as a holdout from the commercial web, Wikipedia has become central to debates over who benefits, and who pays, as AI companies scrape the web to train large language models.
Aggressive data collection from sites like Wikipedia has raised concerns about whether nonprofit platforms are indirectly subsidizing the AI boom.
The foundation had already begun formalizing such relationships in recent years, signing Google as one of its first enterprise customers in 2022, followed by agreements with smaller AI-focused companies like search engine Ecosia last year.
While AI training has triggered lawsuits and copyright disputes elsewhere, Wikipedia founder Jimmy Wales has taken a more pragmatic stance.
“I’m very happy personally that AI models are training on Wikipedia data because it’s human-curated,” Wales told The Associated Press. “I wouldn’t really want to use an AI that’s trained only on X, you know, like a very angry AI,” he added, referring to Elon Musk’s social media platform.
Wales said Wikipedia does not want to block AI companies but expects them to contribute financially. “You should probably chip in and pay for your fair share of the cost that you’re putting on us,” he said.
The push comes amid shifting online behaviour. The Wikimedia Foundation last year reported an 8% drop in human traffic, even as visits from bots, often disguising themselves to evade detection, surged and placed heavy strain on Wikipedia’s servers. At the same time, AI-powered search overviews and chatbots increasingly summarise information directly, rather than sending users to websites via links.
Wikipedia remains the ninth most visited site globally, hosting more than 65 million articles in over 300 languages, edited by around 250,000 volunteers. Its funding still largely depends on about 8 million donors, most of them individuals.
“But our infrastructure is not free,” Wikimedia Foundation CEO Maryana Iskander said in a separate interview with AP in Johannesburg. Maintaining servers and systems that allow both people and tech companies to draw data from Wikipedia comes at a cost, she noted. Iskander, who is stepping down on January 20, will be succeeded by Bernadette Meehan.
Wales was blunt about donor intent. “They’re not donating in order to subsidize these huge AI companies,” he said. “They’re saying, ‘You can’t just smash our website. You have to come in the right way.’”
Beyond monetisation, the foundation also sees potential upside from AI. Its AI strategy includes tools to reduce repetitive work for editors, such as automatically fixing dead links, and possibly transforming Wikipedia’s search experience into a more conversational, chatbot-like interface that directly cites articles.
