Skip to content
AIGP

RFC-025: Cognitive Harm Governance — Psychological Safety, Reality Orientation, and Human Agency in AIGP — 3. Design Thesis

AIGP Specification › RFC-025: Cognitive Harm Governance — Psychological Safety, Reality Orientation, and Human Agency in AIGP › 3. Design Thesis

← 2. Motivation · Section index · 4. Scope →

3. Design Thesis

Cognitive Harm Governance is based on ten claims:

  1. Human harm includes cognitive and psychological harm.
  2. Cognitive harm may emerge over interaction trajectories, not only isolated outputs.
  3. AI systems should not reinforce delusion, paranoia, dependency, self-harm, coercive attachment, or isolation.
  4. AI systems should preserve reality orientation, human agency, dignity, and access to human support.
  5. Psychological harm governance must avoid stigmatizing the user.
  6. The AI system should not claim clinical authority unless properly authorized and bounded.
  7. Cognitive safety must be expressed as Governance Intent.
  8. Cognitive safety constraints must be machine-verifiable where possible.
  9. Cognitive safety interventions must be evidenced through D-DNA.
  10. Replay must reconstruct the interaction trajectory without exposing more sensitive personal content than necessary.

Doctrine:

Human harm includes cognitive harm.
Cognitive harm is temporal.
Reality orientation is a protected outcome.
Human agency is non-delegable.
D-DNA must evidence the safety trajectory.

← 2. Motivation · Section index · 4. Scope →