The Liar's Dividend Without a Liar: Provenance Doubt as an Ambient Cost to Brands That Did Nothing

Abstract

Authentic material loses credibility when audiences are made to suspect it. Warnings wrongly applied to true news reduce its credibility (Freeze et al., 2021). Deepfake labels attached to genuine political video depress perceived credibility for days (Hameleers & Marquart, 2023). Politicians who falsely cry misinformation retain support that an apology would not have preserved (Schiff, Schiff, & Bueno, 2024), the effect Chesney and Citron (2019) named the liar's dividend. Each of these requires a trigger: a label, an accusation, a denial. Someone acts, and credibility falls. The untriggered erosion itself was anticipated — awareness alone primes doubt (Chesney & Citron, 2019) — and has now been measured: in the largest audio deepfake listening study to date, accuracy on authentic recordings fell from 72.7% to 64.1% against a 2021 baseline while detection of fakes held stable (Müller & Choong, 2026). A commercial penalty is likewise documented: authentic pre-2021 advertisements draw the same AI suspicion as generated ones (Ipsos & Newhouse School of Public Communications, 2026). What that literature does not supply is a theory of the brand asset the penalty depletes, or of what defends it. This paper concerns that untriggered case where the cost falls on a brand. A luxury house publishes a campaign. Nobody labels it, accuses it, or denies anything about it. If audiences nonetheless discount it because synthesis has become possible, the cost falls on a brand that did nothing, imposed by no one. We call this the ambient provenance discount and argue that it is the form the phenomenon takes in commercial communication, where there is no defendant and no scandal. We develop a framework for what such a discount destroys. Indexical capital is the accumulated, verifiable evidence that human hands and judgement touched a brand's artefacts. It is cheap to claim and expensive to prove, and generative capability attacks the proving. The framework's central distinction is between depicted eventhood — visible audience, apparent contingency, properties of the image, and therefore generable — and verified eventhood — a dated public calendar, independent press physically present, uncoordinated third-party recordings, a network external to the image. We argue that only the second can defend indexical capital. An exploratory pilot on 6,079 comments from 30 luxury fashion films, published before generative photorealism and read individually where triage surfaced them, finds synthetic attribution essentially absent while production is discussed at a substantially higher rate and reality is occasionally questioned — of the depicted scene rather than of the image. A protocol for a within-house comparison and an experiment isolating verified eventhood follow.