Keep an AI-use ledger for every Short you upload (Python)

Reports on YouTube's AI-labeled content (HeyOrca, read 2026-10-07) keep coming up in creator threads. This post does not interpret YouTube's disclosure rules; read the Help Center text for that. It covers the part you control: knowing, per upload, which synthetic steps were used.

A ledger is one row per Short with a column per step. It also gives you a cost per video.

The ledger

The prices are the repo's billable figures, read 2026-10-07. Edit the dictionary when your catalog changes; the catalog endpoint returns live prices.

import csv, sys
COST = {"tts": 0.0285, "music": 0.125, "timeline": 0.10,
        "captions": 0.20, "image": 0.04375}
shorts = [
    ("reply-001", ["image", "tts", "timeline"]),
    ("demo-002", ["image", "tts", "music", "timeline", "captions"]),
]
w = csv.writer(sys.stdout)
w.writerow(["short", *COST, "total"])
for name, steps in shorts:
    row = [1 if k in steps else 0 for k in COST]
    total = sum(COST[k] for k in steps)
    w.writerow([name, *row, round(total, 5)])

Using it

Add a column for 'real person on camera' and 'real voice' too. The columns that matter for a disclosure decision are the ones YouTube asks about, so copy them from the current form rather than from this post.