See It Work · S2 Vol 1 · Sovereign Inference & Memory · Chapter 3

What a tamper-proof receipt actually looks like

A receipt here isn't a log line you have to trust — it's evidence. Nine facts about one piece of AI work, locked together with a single digital signature (a tamper-proof seal made with a private key only your system holds) and linked to the record before it. Change any one fact and the seal no longer matches — so tampering shows.

What a tamper-proof receipt actually looks like — full detailed chart

The full detailed chart. Condensed for print legibility in the book; shown here at full size.

This is the moment 'this AI work happened' stops being something you assert and becomes something you can hand an auditor — and they can check it without trusting you at all.
Your machine · one commandready

The command produces a receipt with this shape — the nine facts that make a piece of AI work provable:

Inference Receipt
which AI ranits signing fingerprint
input & outputtamper-proof fingerprints
sensitivity labelGREEN / YELLOW / RED
linked to the record before itso the chain can't be faked
one sealcovers all nine facts

Every piece of AI work your operation runs can produce this same shape — which is how months of activity become a chain an auditor can walk end to end.

For the technical reader — the command, and how to verify it yourself
# one line · you do not need to run this
bl-test
bl-test
# -> a sealed 9-fact receipt linked into the chain

Full step-by-step is in Appendix RX: Hands-On Demonstrations in the book.

← All walkthroughsNext: Chapter 4 →