Prompt Five Builds An AI Production Line. None Of The Seven Mentions Disclosure.

Seven prompts, posted one per comment, for building a YouTube channel from zero on a 90-day plan. They are better written than most — specific, bracketed, and carrying guards that a lot of growth prompts leave out.

Five of the seven are strategy and analysis. One is monetisation. And one, number five, is a factory:

“Act as a YouTube production manager. Design an AI-assisted workflow for [CONTENT TYPE] using [MY TOOLS]. Cover research, topic validation, scripting, voiceover, recording, editing, thumbnails, publishing, analytics, and repurposing into Shorts and other platforms. Include practical workflows, reusable templates, quality checks, and clear points where human review is required.”

That is a production line that runs from idea to published Short, with AI at the voiceover and editing stages. It is also the one part of this card that touches a written YouTube rule — and none of the seven prompts mentions it.

The rule prompt five lands on

YouTube requires creators to disclose realistic altered or synthetic content when they upload. The requirement is narrow and specific, which is what makes it usable. From the help page, quoted exactly:

Disclosure is required for content that “makes a real person appear to say or do something they didn’t do”, that “alters footage of a real event or place”, that “generates a realistic scene that didn’t actually occur”, or that “creates music that’s the main focus of the video”.

It is not required for non-realistic AI content, or for edits that are “primarily aesthetic” — the page names beauty filters, colour adjustments and caption creation.

And: creators who consistently do not disclose “may be subject to manual application of a label” and may face “penalties from YouTube, including removal of content or suspension from the YouTube Partner Program”.

Now run prompt five’s pipeline against that.

Notice how much of that table says no. This is not a warning that AI-assisted video is against the rules — most of prompt five’s pipeline is nowhere near the requirement. It is that the one question you have to answer at upload time is not in the prompt that designs the upload process.

There is a second policy in the same neighbourhood, and we flag it with less confidence because we could not reach the primary page. In July 2025 YouTube updated its Partner Programme guidance on mass-produced and repetitious content. Reporting of YouTube’s own statements is consistent on two points: it is not a ban on AI-generated or reused video, provided the work adds original value such as commentary, editing or narration; and enforcement against repetitive, spam-like output was to increase. A prompt that asks for “reusable templates” and “repurposing into Shorts” is building exactly the kind of pipeline where that line gets tested.

The card has three guards. All three are about taste.

This is what makes the omission interesting rather than careless. The author clearly thinks about restraint — there are three explicit guards across the seven prompts, and most cards of this type have none.

That third guard is the place to put the fix, because the prompt is already asking for it. One clause:

“…and clear points where human review is required, including a disclosure check at upload: for each output, state whether it shows a realistic scene, person or event that did not occur, and therefore whether YouTube’s altered-or-synthetic-content disclosure applies.”

That turns the question from something you remember into something the workflow produces. Which is the whole point of prompt five.

What the other six need that the model has not got

Briefly, because this is the familiar part. Three of the seven ask for work built on numbers that live in YouTube Studio, not in a model.

One structural note

Three of the seven are review loops at different cadences: prompt 1 sets a 90-day plan with “a weekly action plan”, prompt 4 asks what to check weekly, and prompt 7 builds “a weekly and monthly review system” with “a simple scorecard I can reuse every week”.

Run all three and you will own three scorecards that nobody reconciles. Run prompt 7 first, keep its scorecard, and let prompts 1 and 4 feed it. The card does not say to do that, and the order it lists them in — plan, position, concepts, analytics, production, money, review — puts the thing that measures everything at the very end. It is the same trap as a prompt that asks for a ninety-day plan and twenty ideas in the same breath: the parts are fine, the arithmetic between them is nobody’s job.

Four honest caveats

1. The disclosure quotes are primary; the July 2025 policy is not. The altered-or-synthetic-content requirements come from YouTube’s own help page. The mass-produced-content update we could only reach through reporting of YouTube’s statements, so we have described it at that confidence.
2. Disclosure rules are narrower than people assume, and they move. Most AI-assisted editing is explicitly outside them. We are flagging a check to build in, not a reason to avoid the tools.
3. We did not run the seven prompts. This is the card read against published policy, not a report on what any model returns.
4. This card is four weeks old. It was posted on 21 September, and thresholds, policies and the prompts’ usefulness all age differently. The strategy prompts age slowest; the monetisation one ages fastest.

The one-line version

Prompt five designs the pipeline that takes a script to a published Short with AI doing the voiceover and the edit, and it already asks for “clear points where human review is required” — it just does not know that YouTube has published one of those points, with penalties attached.

Add the disclosure check to prompt five. Paste your real numbers into prompts 2, 4 and 6. Build prompt 7’s scorecard first.