Wikimedia complains about AI bots scraping as it strains its servers, causing bandwidth to surge by 50%. The Wikimedia Foundation reports ...
 |
| Wikimedia complains about AI bots scraping as it strains its servers, causing bandwidth to surge by 50%. |
The drop may seem small, but at Wikipedia’s scale — with over 20 billion monthly page views — even a single-digit percentage means hundreds of millions fewer visits. This shift reflects a major behavioral change: users increasingly get information from AI-driven summaries rather than clicking through to source material.
A Subtle but Serious Decline
Platforms like ChatGPT, Google Search Generative Experience (SGE), and Perplexity AI aggregate Wikipedia’s data to form concise AI-generated answers — effectively transforming the way people consume knowledge. While this benefits users seeking quick results, it weakens the traffic loop that sustains collaborative platforms like Wikipedia.
Wikimedia now believes that AI-generated answers, not declining interest in knowledge, are responsible for the dip. The foundation warns that this trend could break the feedback cycle that keeps information accurate and up to date.
Wikipedia’s open model depends on two things: readers who visit and editors who contribute. When fewer people visit, fewer are inspired to edit, correct, or update information. This creates a cascading effect that could degrade both the speed and quality of updates across the platform. A decline in user engagement also directly impacts Wikimedia’s fundraising — a critical source of income that keeps Wikipedia ad-free. “Generative AI tools rely on Wikipedia, but they also risk starving it,” said a Wikimedia representative. “If users never return to read or contribute, the knowledge ecosystem that supports AI itself will begin to erode.”
In a public statement, Wikimedia urged AI developers and search engine companies to include clearer source attribution in AI-generated content. It wants LLMs and search systems to feature prominent, clickable citations that lead users back to original Wikipedia pages.
OpenAI, Google, and Anthropic already have partnerships with Wikimedia through the Wikimedia Enterprise API, but the foundation believes more transparency is needed — especially as AI systems become the default way people discover information online.
Experts say the issue reflects a larger tension in the digital ecosystem: AI models depend heavily on open data, but open data projects depend on human traffic and engagement to survive. Without that cycle, free knowledge initiatives like Wikipedia, Stack Overflow, and academic archives could struggle to maintain their quality.
“It’s a paradox,” said Dr. Emily Clarke, a digital knowledge researcher at the University of Oxford. “AI models are parasitic on open information, yet their success may undermine the very systems that produce that information.”
Wikimedia is now testing new tracking tools to better distinguish between human and AI-driven traffic. It’s also exploring partnerships with major LLM developers to ensure Wikipedia content is used responsibly and credited visibly in AI-generated answers.
Some experts believe this is the start of a new phase for the web — one where information lives inside AI systems rather than on traditional websites. Others see this as a call to defend the principles of open access and collective contribution before they fade away behind algorithmic summaries.
The foundation’s message is clear: as AI becomes the interface to knowledge, transparency and attribution must evolve — or the internet’s most valuable commons could quietly fade into the background.
The Future of Knowledge Depends on Collaboration
Wikimedia’s findings underline an emerging truth: the future of knowledge-sharing cannot be sustained by automation alone. It requires a global community of contributors, transparency from AI companies, and a renewed commitment to open, verifiable information.
In a world where AI answers everything, it’s vital to remember where those answers come from — and to support the people and platforms that make them possible.
