Wikipedia is working with major technology companies on licensing agreements to offset the costs of high-volume access used for training AI models. Co-founder Jimmy Wales said at the Reuters NEXT summit that AI firms’ automated scraping of Wikipedia content drives significant expenses for servers, RAM, and caching.
The nonprofit Wikimedia Foundation already has a deal with Alphabet’s Google, signed in 2022, and is in talks with other AI companies, including OpenAI and Meta Platforms. Wales emphasized that public donations, which support Wikipedia, are not intended to subsidize the development of commercial AI products.
The initiative raises broader questions about whether for-profit AI companies should compensate public and nonprofit content providers for datasets that power their models. Wales noted potential technical measures, such as AI Crawl Control, to limit bot access, while acknowledging the tension with Wikipedia’s commitment to open knowledge.
Wikipedia has operated as a nonprofit for over two decades, relying on volunteers and donations. Wales said maintaining neutrality remains challenging but essential, especially during global conflicts, as the platform balances accessibility with operational sustainability.
