Synthetic Media & Provenance · Capstone brief

PROV-08 · Political ad case study

Take one real political campaign's approach to AI disclosure — what it did, what it was required to do, and what it said about it — and assess it against a standard you state first. Where the campaign disclosed nothing, that absence is the case.

The question

Take one real political campaign's approach to AI disclosure — what it did, what it was required to do, and what it said about it — and assess it against a standard you state first. Where the campaign disclosed nothing, that absence is the case.

System / materials

Publicly available campaign material and public reporting, cited with dates. The disclosure requirements that actually applied — platform policies, and any statute or regulation in force for that race and jurisdiction, read at the source rather than from commentary. A written assessment standard, fixed before the analysis. This is analysis of published material: no synthetic political media is produced, and no candidate's likeness or voice is synthesized for any purpose.

Expected failure modes

Choosing a case for its partisan valence rather than its disclosure question, which makes the analysis unreadable to half the audience. Assuming a rule applied without checking whether it was in force in that jurisdiction at that time. Confusing a platform policy with law. Presenting an inference about intent as a finding. Grading a campaign against a standard invented after seeing what it did.

Done looks like

A case assessment: the campaign material described factually with dates and sources; the requirements that actually applied, cited; the pre-stated standard and the assessment against it, separating what is documented from what is inferred; a note on what a reader could and could not have known at the time; and a proposed norm that would have improved the situation, with its enforcement path.

Five C's

CT: separating documented fact from inference about motive. CR: choosing a case for its disclosure question. CO: a peer with a different political view reads the analysis and reports whether it reads as fair. CM: an assessment a general reader trusts. CZ: the voter making a decision on what they saw.

Mentor role

A political reporter, standards editor, or government teacher reviews the standard before the analysis. Standing instruction: reject a case chosen for partisan advantage, and reject any inference about intent presented as fact. School-supervised.

Rubric calibration

R1: one campaign, one race, one standard. R2: every claim cited with a date. R3: comparator is the requirement in force. R4: inference labeled as inference throughout. R5: reads as fair to a reader who disagrees. R6: refuses claims about intent the record cannot support.

Two ways this goes wrong

(a) An argument about a candidate wearing a disclosure analysis as a costume. (b) A rule cited that was not in force in that jurisdiction at that time.

Credit lane fit

Lane A immediately (journalism, government, or AP Research). No verified credit claim.