.
Image: FC
The Wikimedia Foundation (WMF), the nonprofit behind the global encyclopedia Wikipedia, has offered up a stark message to generative-AI firms: quit scrapping the freely accessible site, and instead pay for access via its enterprise API.
In a blog post published this week, WMF said that AI companies should engage with Wikipedia’s content responsibly, meaning both proper attribution of the human contributors and use of the paid enterprise API product rather than unstructured scraping of its pages.
The backdrop is a shifting web-landscape. User traffic to Wikipedia has dropped; the Foundation reported an 8 % year-over-year decline in “human page views.”
According to TechCrunch, WMF did note a surge in traffic in May and June that was driven by AI bots scraping the site while attempting to pass as humans.
For Wikimedia, the combination of declining visits and rising scraping activity threatens both the volunteer model (fewer people visiting means fewer contributors) and the financial sustainability of the organisation (fewer donations and less engagement).
By shifting major AI customers into the paid enterprise API lane, Wikimedia aims to stabilise its business model and protect its infrastructure from heavy loads.
WMF has also asked AI companies to clearly attribute Wikipedia’s human editors when their content is used in AI outputs.
The company noted that fewer visits to Wikipedia could translate into fewer volunteer contributors and fewer donors, thereby potentially weakening the resource that AI firms rely on.
For large language model (LLM) operators and other generative-AI companies, the ask from Wikimedia could trigger material changes: reconsidering how they access reference knowledge, assigning budget for API access rather than relying on free web scraping, and re-thinking attribution practices.
On the other hand, the lack of explicit enforcement (there’s no mention of legal threats or penalties in the blog post) suggests this is a plea rather than a crackdown, at least for now.
For Wikipedia, this seems like an attempt to adapt its nonprofit, volunteer governance model to the realities of an AI-driven knowledge economy.
Earlier this fall, Elon Musk’s AI company xAI launched Grokipedia, an AI‐powered encyclopedia intended to rival Wikipedia.
Grokipedia went live with about 885,000 articles (now 885,279), compared to Wikipedia’s 8+ million English-language entries at that time.
The site uses the same large language-model tech that powers Musk’s chatbot on platform X.
Musk has framed the move as challenging Wikipedia’s volunteer-editor model and alleged “liberal bias,” pitching Grokipedia as a “truth-oriented” knowledge engine driven by AI rather than consensus editing.
The Wikimedia Foundation’s push to monetise and control API access comes at a moment when alternative knowledge systems like Grokipedia are emerging.
If Wikipedia becomes more gated (via paid API access) for AI firms, emerging rivals might fill the knowledge gap, for better or worse.
The philosophical underpinnings differ: Wikipedia emphasises open, volunteer-driven governance, whereas Grokipedia emphasises machine-driven output and centralised control by Musk’s ecosystem.
From a business standpoint, the scramble for “trusted” data sources for generative AI may heighten the value of sanctioned APIs like Wikimedia’s, or, conversely, increase incentives to bypass them.
FAST COMPANY (SA)