YouTube Shorts AI slop: an originality checklist for your pipeline

Search suggestions today include 'youtube shorts ai slop'. This post does not claim to know how YouTube ranks or demotes anything; HeyOrca (read 2026-10-07) reported ML comment filtering and AI labeling work, not ranking changes. It is a checklist for making something specific on purpose.

Checklist

Each item names the Sume step that supports it, with its price from the repo (read 2026-10-07).

• Write the script yourself or edit it hard. A voice at $0.0475 per 1,000 characters is cheap, so the script is where your effort goes.
• Use your own footage or stills as the base and generate only what you cannot shoot.
• Pick one voice and keep it across episodes so viewers know the series.
• Set captions to a style you reuse (video captions, $0.20 per job).
• Compose a music bed for the series ($0.125) and reuse the prompt wording.
• Keep the AI-use ledger so the label question has an answer.
• Watch the first 3 seconds. If the opener could belong to any channel, rewrite it.

What this costs

A reply-style Short is $0.158 and a captioned product Short is $0.6675. Money is not the filter; taste is. Spend the saved time on the script.