Compliance

The same rule set runs in the first-cut wizard, in Studio, on a Canvas video node, and on the publish queue. A live session uses it too. There is no switch that turns a block off.

What a block looks like

• You see the rule name, the reason, and a concrete fix. You do not see an error code.
• Generation of a blocked script does not start. No credits are charged for that attempt.
• A warning, such as an absolute claim, is shown and does not stop generation.
• Publish still refuses a block even when the video already exists.
• AI disclosure starts checked on the composer. Turn it off only when a person filmed the cut.
• If a generation is charged and the provider then fails, the original charge is refunded. A block never reaches that path.

The rules in this product

• Medical or therapeutic claims

  Block. Describe the product's ordinary function instead.

• Earnings promises

  Block. Remove any promise of profit, income, or returns.

• Prohibited product categories

  Block. The content cannot be generated, published, or streamed.

• Absolute or unverifiable claims

  Warning. Replace them with wording you can stand behind.

• Missing AI disclosure on a synthetic presenter

  Warning. Keep an on-screen disclosure for the full duration and mention it in the description.

• Unattended digital-human selling

  Block on a live script unless a person attests they are watching the stream. AI-assisted selling is allowed. Replacing the seller is not.

What this check is not

• It is a product check, not legal advice, and not a promise that TikTok or YouTube will approve the post.
• It reads the text you type. It does not scan an uploaded face, and there is no face-identification provider in this product.
• It does not estimate sales, GMV, or a market rank.
• The examples in the composer are labeled samples, not customer content.
• Automatic wording detection currently covers Chinese, English, and Japanese only. Medical claims written in Korean, Spanish, German, or French are not scanned yet.