Wikipedia says traffic is falling due to AI search summaries and social video (techcrunch.com)

The Wikimedia Foundation reports an 8% year‑over‑year drop in human pageviews after an update to its bot‑detection systems revealed that much of an earlier spike was driven by bots engineered to evade detection. Wikimedia’s Marshall Miller attributes the broader decline to changing information habits: search engines increasingly surface generative‑AI summaries instead of linking to source pages, and younger users are turning to social video platforms for answers. (Google has disputed the claim that AI summaries reduce traffic.) Wikipedia even experimented with its own AI summaries but paused the effort after editor pushback. For the AI/ML community this trend matters both technically and ethically. Reduced direct visits threaten the volunteer and donor ecosystems that curate the encyclopedia—undermining the human‑authored training data many models rely on—and highlight urgent issues around attribution and provenance when models or platforms reuse community‑created content. Wikimedia is building an attribution framework, has two teams focused on audience outreach, and is calling on platforms that repurpose Wikipedia content to drive readers back to the site. The foundation’s plea—click through citations, support content creators, and prioritize source transparency—underscores the need for ML systems and search interfaces to surface provenance so users and contributors remain connected to the original knowledge creators.