The AI Perception System can categorize perception targets into friendly, neutral and hostile. This is useful for filtering out what you want certain AI’s to detect, and for improving performance to avoid running checks against unwanted targets.
Detection by affiliation requires implementing IGenericTeamAgentInterface on the actors that are being perceived (typically a Pawn of some variety), or on the actor doing the perceiving (typically an AIController).
See IGenericTeamAgentInterface for implementation information.