Wikipedia, long hailed as one of the last bastions of reliable, community-curated knowledge on an increasingly commercial and AI-saturated internet, is seeing a noticeable dip in human engagement.
According to new data shared by Marshall Miller of the Wikimedia Foundation, human pageviews on Wikipedia have fallen by about 8% year-over-year.
Miller, who detailed the findings in a new blog post, said the decline became apparent after an update to Wikipedia's bot detection systems revealed that a large portion of the site's unusually high traffic between May and June actually came from bots designed to avoid detection. The adjusted data, he said, now paints a more accurate picture -- one that shows a steady decline in organic, human-driven visits to the encyclopedia.
The explanation for this trend lies in the broader shifts shaping how people consume information online. Miller identified "the impact of generative AI and social media on how people seek information" as key factors behind the decline. Search engines, including Google, are increasingly providing direct, AI-generated answers to user queries rather than linking to source websites like Wikipedia. This means that while Wikipedia content continues to inform millions of users daily, many of those users never actually click through to the site.
At the same time, younger generations are gravitating toward short-form video platforms like TikTok, Instagram Reels, and YouTube Shorts for quick, visually engaging answers to their questions -- a habit that bypasses the traditional open-web model that Wikipedia helped define.
"Younger generations are seeking information on social video platforms rather than the open web," Miller wrote.
Google, for its part, has disputed claims that its AI summaries or "overviews" in search reduce traffic to knowledge sources. The company has argued that links to external websites are still provided and that users can click through to learn more. But data across the web suggests that the introduction of AI-generated search summaries has indeed shifted user behavior.
However, Miller maintained that Wikipedia remains a critical foundation of the global information ecosystem. Even if readers are not visiting the site directly, the encyclopedia's content still forms the backbone of many AI and search tools.
"Knowledge sourced from Wikipedia is still reaching people even if they don't visit the website," Miller noted.
The challenge, however, is that fewer visits mean fewer opportunities for volunteers to edit and expand Wikipedia's content -- the lifeblood of the platform. It could also translate to fewer small donations, which are essential to funding the Wikimedia Foundation's operations.
"With fewer visits to Wikipedia, fewer volunteers may grow and enrich the content, and fewer individual donors may support this work," Miller cautioned.
The risk is not merely financial or operational. As generative AI systems rely on vast repositories of existing information to generate new text, a decline in Wikipedia's volunteer base could have downstream effects on the quality of knowledge available across the internet. If fewer human editors are updating and correcting Wikipedia, the accuracy of the AI systems that depend on it could suffer as well.
Interestingly, Wikipedia itself experimented with integrating generative AI, testing automated summaries of entries earlier this year. However, the initiative was paused after editors voiced concerns that AI-generated text could compromise the platform's integrity, transparency, and editorial standards.
In response to the decline in engagement, the Wikimedia Foundation is taking several steps to ensure Wikipedia remains relevant and visible in a changing information landscape. The organization is developing a new framework for attributing content used by AI models and search engines, aiming to reinforce the visibility of Wikipedia's role in shaping digital knowledge. It also has two dedicated teams working to expand the platform's reach among new readers globally, particularly in emerging markets where mobile and social platforms dominate.
Miller's post also included a broader appeal to readers and internet users to support what he called "content integrity and content creation." He urged users to look beyond AI summaries and social media snippets by clicking through to original sources, checking citations, and contributing where possible.
"Talk with the people you know about the importance of trusted, human-curated knowledge," he wrote, "and help them understand that the content underlying generative AI was created by real people who deserve their support."
Since its launch in 2001, Wikipedia has grown into one of the most visited websites in the world and one of the few remaining large-scale collaborative projects run by volunteers. Its open-access model has stood in contrast to the algorithm-driven, ad-funded structures that dominate most of today's internet. Yet even Wikipedia is now facing the same pressures as other legacy web institutions -- declining visibility, competition from AI, and changing consumption habits that favor speed over depth.
The recent revelation that human traffic is falling may not mean Wikipedia's influence is waning, but it underscores a pivotal moment. If the encyclopedia's role as a public good is to endure in the age of generative AI, it will need both institutional adaptation and renewed public engagement.