Skip to content
AIGP

RFC-025: Cognitive Harm Governance — Psychological Safety, Reality Orientation, and Human Agency in AIGP — 32. Summary

AIGP Specification › RFC-025: Cognitive Harm Governance — Psychological Safety, Reality Orientation, and Human Agency in AIGP › 32. Summary

← 31. Example Scenarios · Section index

32. Summary

Cognitive Harm Governance establishes that:

human harm includes cognitive and psychological harm
harm may emerge over interaction trajectories
AI systems must preserve reality orientation and agency
policy prose is insufficient
cognitive safety must be expressed as Governance Intent
SGL constraints must inhibit harmful interaction patterns
ENFORCE may activate cognitive safety controls
VGL may project cognitive state only with privacy safeguards
D-DNA must evidence the trajectory, constraint, intervention, and replay

The final principle is:

AI systems that influence human belief, emotion, agency, or reality orientation must be governed not only by content rules, but by trajectory-aware cognitive safety constraints that are privacy-preserving, human-centered, and D-DNA replayable.

← 31. Example Scenarios · Section index