Add a safeguarding review before approving an AI brief involving children, whether the proposed child is identifiable or fictional. Platform moderation is not evidence that guardian authority, dignity or the intended distribution has been checked.
Start with necessity, not visual appeal
Ask why a child needs to appear. Could an object, animation, adult explanation or text convey the same point without exposing or implying a child's personal circumstances?
For school, tuition and family-oriented campaigns in India, pay particular attention to uniforms, school names, locations and language-specific captions. Review whether their combination could identify a real child or suggest a real institution's involvement. Do not supply actual pupil records, family communications or private photographs as example inputs.
Use a guardian-authority and dignity checklist
Record a clear answer and review owner for each question:
- Authority: If a real child is involved, what evidence establishes who may approve the proposed use?
- Scope: Does the approval being reviewed address synthetic alteration, campaign purpose and intended distribution?
- Child perspective: Has the review considered the child's understanding and comfort where appropriate?
- Dignity: Does the scene humiliate, stereotype or pressure a child, or attach an invented personal difficulty to them?
- Identification: Are names, badges, addresses or distinctive surroundings unnecessary?
- Future context: Could a clipped version create a damaging impression later?
- Withdrawal planning: Who handles concerns and pauses further distribution?
These questions are a safeguarding workflow, not a statement of legal requirements or provider-specific rules.
Hypothetical example: a tuition campaign
Imagine a fictional tuition provider proposing a 20-second advertisement showing an identifiable pupil looking distressed beside a failing mark sheet. Its inputs are a proposed pupil photograph, invented marks and a claim that the course transformed the child's confidence.
Do not upload those materials. The brief attaches invented academic and emotional circumstances to a real child. Guardian approval alone would not answer whether that portrayal is dignified or misleading.
A revised concept could show an anonymous desk, fictional practice questions and an adult explaining the study process. Evaluate whether the message remains useful without inventing a child's failure or success. If it does, the lower-exposure approach meets the communication need.
Set a safeguarding stop point
Stop when authority is unresolved, identifying details are unnecessary, or the persuasive idea depends on embarrassment. Next, ask a designated safeguarding reviewer to assess a non-identifying storyboard before production.
If sourcing production access, review the AI tools catalog, checking access format and saving terms before buying. Marketplace account/code/token offers are not proof of official subscription equivalence or safeguarding approval.
Sources and scope
Source material checked 29 October 2026; features/policies may change. Runway's Usage Policy establishes automated and internal human review for harmful inputs and outputs. The supplied evidence does not specify guardian-verification procedures or detailed child-depiction rules. Consult the current policy; the checklist and hypothetical revision are editorial safeguarding advice.