Wikimedia warns AI-generated answers are hurting Wikipedia traffic

Wikimedia is raising concerns that the rise of large language models (LLMs), AI-generated summaries and chatbots is reducing direct visits to Wikipedia — a core source of verifiable online knowledge. In a recent blog post, Marshall Miller, Wikimedia’s senior director of product, links an observed 8% year-over-year drop in page views to generative AI and changes in how search engines present answers.

Wikimedia reports that improving bot detection changed its metrics and exposed a sharper decline in human page views. The foundation argues that when search engines and chatbots provide direct answers based on Wikipedia content, users may never visit the original articles — undermining volunteer engagement, funding and the site’s long-term reliability.

Key points:

• Wikimedia recorded an 8% YoY drop in page views after refining bot detection.
• The foundation attributes much of the decline to AI chatbots and AI-generated summaries appearing in search results.
• Wikimedia warns fewer visits could lead to fewer volunteers and reduced funding, threatening content quality and verifiability.

Wikimedia proposes that LLMs and search platforms be more intentional about surfacing sources and giving users clear opportunities to visit and interact with original content. Earlier this year, Wikimedia considered offering AI-generated article summaries on its pages but dropped the project after pushback from volunteer editors.

For more background, see the original coverage on Engadget or read Wikimedia’s blog at blog.wikimedia.org.

Discussion: Do you think AI platforms should be required to link back to source articles like Wikipedia so readers can verify and contribute to the originals?