Wikimedia says AI bots and summaries are hurting Wikipedia's traffic (www.engadget.com)

Wikimedia — via a blog post from product lead Marshall Miller — warns that generative AI is siphoning traffic and attention from Wikipedia. After tightening bot-detection to better separate automated crawlers from human readers, the foundation found an 8% year‑over‑year drop in page views it attributes to LLM chatbots and AI-generated summaries in search results that deliver direct answers often based on Wikipedia content. Wikimedia also calls out increasingly sophisticated bot crawlers that scrape and republish material, complicating traffic measurement and content stewardship. Earlier this summer the foundation considered deploying its own AI summaries at the top of articles but shelved the plan after pushback from volunteer editors. The significance for the AI/ML community is twofold: first, a feedback loop where model providers and search engines use Wikipedia as a knowledge source while reducing visits undermines the very resource that sustains verifiable, neutral content; second, it highlights practical issues around provenance, attribution and the economics of open knowledge. Wikimedia urges LLM and search providers to surface sources, link back to originals, and elevate opportunities for users to visit and participate in source communities — a technical and policy ask that affects dataset curation, model evaluation, and long‑term sustainability of web knowledge ecosystems.