Why Wikipedia’s Page Views Fell 8%—and What AI Has to Do With It

Wikipedia’s human page views fell roughly 8% year over year in the period Wikimedia examined. CEO Bernadette Meehan points to a shift in how people find information: AI tools and search summaries can deliver answers without sending readers to the encyclopedia. Wikimedia also identifies social media as part of the changing landscape, so the figure should not be treated as proof that AI alone caused the decline. 19

Why Wikipedia is seeing fewer visits

Wikimedia reported the drop after updating how it detects automated traffic. Its analysis found that some unusually high traffic had come from bots designed to evade detection, while human page views had declined compared with the same months a year earlier. 914

The Foundation says the shift also reflects how people now seek information: generative AI and social platforms, along with search engines that answer questions directly, can reduce the need to click through to a Wikipedia article. The reported 8% comparison is a measure of traffic, not a breakdown of how much each trend contributed. 9

That distinction matters. Meehan describes AI-driven changes as a concern, but the available reporting does not establish what portion of the decline is attributable to generative AI specifically. 19

AI uses Wikipedia content, but fewer users may arrive at the site

Wikipedia presents a tension in the AI era: its human-curated information is useful to AI systems, while AI-generated answers can give users information without a visit to Wikipedia. Wikimedia says that large-scale commercial access can also create infrastructure costs. 117

The Foundation offers Wikimedia Enterprise, a paid service for reliable, high-volume access to its data. Axios reports that publicly disclosed enterprise customers’ payments account for 10% of the Foundation’s operating expenses. This is a commercial access route for organizations using content at scale—not a charge for people reading Wikipedia normally. 1

Donations and volunteer editors are a risk, not a measured 8% decline

Fewer visits could mean fewer chances for readers to donate or discover how to contribute as editors. That is a potential concern for a nonprofit whose primary income includes public donations and whose articles depend on volunteer work. But the reported 8% figure measures page views; it does not show an equivalent fall in donations or volunteer recruitment. 217

The longer-term issue is whether the knowledge ecosystem can remain sustainable when AI products use Wikipedia-derived information while sending fewer people to the source. Meehan’s argument, as reported by Axios, is that companies benefiting from high-volume access should help support the resource. 1

Meehan’s view: AI should support human editors

Meehan has described a human-first approach to AI: use the technology to support editors while keeping people central to creating and maintaining reliable knowledge. The Wikimedia Signpost interview also notes concerns that machine-generated content can spread disinformation, create security risks, and erode trust. 6

That approach distinguishes assistance from handing over editorial responsibility. Wikipedia’s continued value depends not just on content being available to AI systems, but on people being able to check, improve, and curate it. 615