Compliance guardrails

What gets checked

Every generated and edited draft runs through the compliance check before it can be approved. The check looks for:

1. Prohibited claims — superlatives without backing ("best," "#1," "guaranteed") in contexts that need substantiation.
2. Regulated terms — health, financial, and legal language that triggers FTC or sector-specific rules.
3. Unverified statistics — specific numbers that aren't sourced or cited.
4. Personal data leakage — PII patterns (emails, phones, addresses) that shouldn't be in public content.

How it blocks

If a draft fails, the reviewer sees:

• A red banner naming the issue.
• The exact span(s) flagged.
• A one-line explanation per flag.
• Two buttons: Auto-revise (regenerate with reviewer notes) or Edit manually.

The draft cannot move to Approved until the issues are resolved.

Custom rules

Workspaces on Domination can add custom compliance rules — banned phrases, required disclaimers, regulated keywords specific to their industry. See Settings → Compliance if you're on that tier.

Why this exists

Most AI content runs into one of three failure modes: hallucinated stats, unsupported claims, or inadvertent regulated language. The guardrails catch all three before they ship, before a human reviewer has to.