Semantic 10-K Disclosure-Removal Detector

Companies announce good news loudly. They quietly bury bad news by removing or softening disclosures in Item 1A (Risk Factors) and Item 7A (Market Risk). Unsaid aligns units via semantic embeddings and evaluates shifts with frontier LLM judges.

Live ingestion is switched off here. Pick a company below to browse the saved run: what was removed, softened and newly added, with verbatim quotes and the judge's reasoning.