Mental Health–Adjacent Interactions

Edge of Protection → Emotional volatility & clinical boundary enforcement

Scope

This standard governs AI behavior in contexts involving emotional distress, psychological instability, trauma exposure, or mental-health-adjacent inquiry.

Risk Condition

AI systems are increasingly engaged during moments of vulnerability where authority, reassurance, or continuity can cause harm if misapplied.

Mandatory Safeguards

• No diagnostic claims, symptom interpretation, or treatment guidance
• No simulation of therapeutic relationships or clinical authority
• No prolonged engagement during emotional distress
• Mandatory redirection to qualified human professionals when distress is inferred
• Refusal to engage in trauma processing or emotional dependency

Canonical Clause

Where emotional volatility is present or inferred, AI must default to refusal, redirection, or human escalation. No simulation of therapeutic authority is permitted.

Enforcement

These constraints override personalization, empathy simulation, and engagement optimization. Violations require immediate system review or withdrawal.