diff --git a/README.md b/README.md
index e5348c3..ed78b17 100644
--- a/README.md
+++ b/README.md
@@ -22,7 +22,8 @@ through afterwards.
 - **ML tagging.** Runs image models in-container to suggest tags, group
   characters, find near-duplicates and power similarity search. Suggestions are
   reviewable — it proposes, you confirm, and it learns which proposals you keep
-  rejecting.
+  rejecting. It ships switched off: turn it on under Settings → System when
+  you want it, and it fetches its model weights then.
 - **Deduplication and provenance.** Everything that arrives is hashed and
   deduplicated by content, metadata sidecars are read wherever the source
   writes them, and every file keeps a record of where it came from.
@@ -119,9 +120,12 @@ needs every credential entered again by hand.
 
 A few other things are worth knowing about the first few minutes:
 
-- **The ML worker downloads its model weights on first boot**, several GB from
-  HuggingFace into `./models`. Until that finishes, tagging is queued rather
-  than broken. It is idempotent — a restart resumes rather than refetches.
+- **ML tagging starts switched off, and nothing is downloaded at boot.** Give
+  the ML lane a slot under **Settings → System** and it fetches its model
+  weights then — several GB from HuggingFace into `./models`, shown as a job
+  under **Settings → Activity** that you can watch and retry. Until it
+  finishes, tagging is queued rather than broken, and the fetch only takes
+  what is missing, so turning the lane off and on again does not refetch.
 - **The gallery starts empty**, and that is the expected state. Add a creator
   under **Subscriptions** and it fills as posts come down.
 - **If you already have a library on disk**, there is no screen that imports