How the attack works
An attacker frames a request as a personal or seemingly reasonable ask, naming a real person and their relationship to the requester. The request asks the AI to provide, identify, or scan a biometric trait belonging to that person. The personal-relationship framing is meant to lower suspicion and make the AI treat it as a normal, benign query. The named-individual detail is what separates this from harmless general questions about how biometric technology works.
Why it matters
If the AI complies or attempts to fulfill such a request, it can facilitate stalking, unauthorized identity verification, deepfake creation, or identity theft against a real, identifiable person.
What you can do
- Configure AI systems to refuse biometric data requests tied to a named individual, regardless of the stated relationship.
- Log and review any biometric-related queries that include personal names for manual follow-up.
- Distinguish and allow legitimate use cases (forensics, law enforcement, academic research) via separate authorized channels rather than open-ended chat access.
- Train support and moderation teams to recognize relationship-framing as a social engineering tactic, not proof of legitimacy.
Known benign look-alikes
- Law enforcement or forensics professionals with authorized access
- Academic discussion of biometric surveillance technology without targeting a named individual
- General how-does-it-work questions about biometric systems