RFC-025: Cognitive Harm Governance — Psychological Safety, Reality Orientation, and Human Agency in AIGP — 13. ENFORCE for Cognitive Safety

AIGP Specification › RFC-025: Cognitive Harm Governance — Psychological Safety, Reality Orientation, and Human Agency in AIGP › 13. ENFORCE for Cognitive Safety

← 12. AIGP-SGL Cognitive Constraints · Section index · 14. AIGP-VGL for Cognitive Safety →

13. ENFORCE for Cognitive Safety

Cognitive harm governance may use ENFORCE Broadcast inside a system or agent orchestration environment.

Example:

enforce_signal:
  signal_id: ebs.cog.000773
  signal_type: ENFORCE
  issuer:
    issuer_id: cognitive_safety_monitor
    issuer_type: runtime_monitor
  ttl_seconds: 600
  scope:
    conversation_session: session.993
    applies_to:
      actors:
        - assistant.companion
        - agent.memory
        - agent.recommendation
  command:
    action: activate_cognitive_safety
    reason: delusion_reinforcement_risk
    required_behavior:
      - prohibit_delusion_validation
      - prohibit_paranoia_reinforcement
      - reality_anchor
      - encourage_human_support
      - avoid_exclusive_attachment_language
      - escalate_if_self_harm_intent
  evidence:
    ddna_required: true

Cognitive ENFORCE may be internal only. It should not necessarily be displayed to the user.

← 12. AIGP-SGL Cognitive Constraints · Section index · 14. AIGP-VGL for Cognitive Safety →