Restricted Prompt Categories

Scenario

Teams receive mixed-quality requests. Some are clearly prohibited, while others need strict review before generation.

Steps

1. Classify request intent.
2. Reject prohibited categories immediately.
3. Rewrite ambiguous prompts with explicit boundaries.
4. Escalate unresolved cases for policy review.

Parameter Guidance

• Reject categories:
  • illegal or harmful requests
  • deceptive impersonation
  • disallowed explicit sexual content
• Restricted categories:
  • real-person identity edits
  • rights-unclear brand requests
  • potentially misleading claims

Failure Troubleshooting

Symptom: Request intent is unclear

Fix:

• Ask for business purpose and ownership context.
• Convert to neutral, policy-safe creative direction.

Symptom: Team disagrees on policy decision

Fix:

• Use written escalation path.
• Record decision and rationale for future consistency.

Compliance Boundary

• Safety policy applies regardless of paid plan or urgency.
• Any request violating law or platform policy must be rejected.

FAQ

Can restricted requests ever be allowed?

Some restricted requests can proceed only after boundary clarification and reviewer approval.

What should happen after a rejection?

Provide a safe alternative prompt path instead of silent failure.