The Trauma-Informed AI Manifesto

Generic large language models are trained to be inquisitive, asking questions like "Can you tell me more about what happened?" or "Describe the event in detail." For a survivor of torture, state terror, or sexual violence, this conversational probing forces them into vivid re-experiencing, inducing acute autonomic panic, flashbacks, and dissociation.

KindMinds adheres to a strict protocol: Never interrogate, never demand forensic narrative, and anchor to the physical room whenever autonomic distress spikes.

Rather than treating every message as flat text, KindMinds deploys real-time function calling to assess nervous system distress (0.0 to 10.0 scale), classifying trauma states into Hyperarousal (fight-or-flight panic), Dissociation (depersonalization/numbing), Intrusive Distress (active flashback intrusion), or Acute Crisis.

This continuous telemetry allows the AI agent to immediately modulate its sentence rhythm, suggest sensory anchors, or trigger crisis escalation protocols before the survivor spirals into complete overwhelm.

Victims of atrocities have experienced the catastrophic loss of bodily and psychological control. Being commanded by an AI—even with good intentions—can evoke feelings of coercion and helplessness.

KindMinds consistently offers collaborative choices: "Would you like us to pause and try a slow breath together, or would you prefer we focus on finding a safe physical place to rest?" The survivor is always the director of their healing journey.

Survivors in conflict zones, oppressive regimes, or domestic violence situations risk retaliation if an abuser or authority discovers their searches. KindMinds operates with zero ad tracking, zero mandatory account creation, and a global double-tap ESC escape shield that wipes active memory and redirects to Google Weather instantly.