RFC-025: Cognitive Harm Governance — Psychological Safety, Reality Orientation, and Human Agency in AIGP — 9. Prohibited Outcomes

AIGP Specification › RFC-025: Cognitive Harm Governance — Psychological Safety, Reality Orientation, and Human Agency in AIGP › 9. Prohibited Outcomes

← 8. Protected Outcomes · Section index · 10. Risk Signals →

9. Prohibited Outcomes

AIGP defines the following prohibited cognitive-harm outcomes.

prohibited_cognitive_outcomes:
  delusion_reinforcement:
    description: AI validates or elaborates a delusional belief as factual.

  paranoia_amplification:
    description: AI strengthens unsupported threat interpretations.

  self_harm_enablement:
    description: AI assists, encourages, normalizes, or operationalizes self-harm.

  coercive_attachment:
    description: AI creates or exploits emotional dependency or exclusive attachment.

  false_clinical_authority:
    description: AI presents itself as a qualified clinician or definitive diagnostic authority when it is not.

  isolation_from_humans:
    description: AI discourages contact with trusted humans or appropriate professionals.

  dependency_escalation:
    description: AI reinforces that the user should rely primarily or exclusively on the AI.

  reality_boundary_erosion:
    description: AI blurs fiction, speculation, roleplay, hallucination, or delusion with reality in harmful contexts.

  compulsive_reassurance_loop:
    description: AI repeatedly provides reassurance in a way that deepens anxiety or compulsive checking.

← 8. Protected Outcomes · Section index · 10. Risk Signals →