Private
Contact and precise location details are redacted.
AI & Safety
Leads are suggestions, never confirmations.
Sensitive details stay restricted and decisions stay human-reviewed.
Locations hiddenHuman reviewNo auto-contact
Explainable
Each lead shows evidence, conflicts, and gaps.
Gated
Review notes do not confirm identity or contact anyone.
Operating rules
Built for cautious use
Responsible AI controls
Lead, not confirmation.
No final AI decisions
No percentages
No auto-contact
Exact locations hidden
Human review required
Failure modes
Reviewer checks
False match
Risk
Families may receive false hope, panic, or misinformation.
Mitigation
Every result is labeled as a lead, not confirmation. The app shows evidence, conflicts, and missing information. Human review is required before action.
Missed match
Risk
A real lead may be overlooked.
Mitigation
Low-confidence leads remain visible, missing fields are shown, and reviewers are encouraged to request more information instead of ignoring uncertain reports.
Privacy exposure
Risk
Survivors, children, or vulnerable people could be located by unsafe actors.
Mitigation
Phone numbers, emails, exact addresses, apartment numbers, and precise locations are redacted or hidden by default.
Bias against names, languages, or descriptions
Risk
People with non-Western names, unclear spellings, multilingual reports, or vague descriptions may be matched poorly.
Mitigation
The app does not rely on name alone. It compares age, broad location, time, clothing, language, and description evidence. It also explains what caused each lead.
Over-reliance on AI
Risk
Volunteers may treat suggestions as truth.
Mitigation
The UI avoids percentages, avoids "confirmed" language, repeats warnings, and requires reviewers to check a human review box.