Safety
Our Safety Commitment
Shadow work surfaces difficult material by design. Every feature in this app is built around the principle that exploration should be challenging, but never harmful.
Safety-First Philosophy
Shadow Self combines deep Jungian psychological exploration with robust automated safeguards. The system operates across four layers, detection, intervention, access, and transparency : designed to be non-intrusive for healthy users while responding quickly when someone needs help.
Real-Time Crisis Detection
Every message to Sage is scanned before normal processing begins.
- • 50+ monitored crisis keywords
- • 3-tier risk levels (low / medium / high)
- • High risk bypasses AI, shows resources immediately
- • Escalation pattern recognition
24/7 Crisis Support Access
Persistent one-tap access to professional crisis support across all screens.
- • 988 Suicide & Crisis Lifeline
- • Crisis Text Line (Text HOME to 741741)
- • One-tap call and text
- • 24-hour dismissal option
Informed Consent System
Content warnings with depth previews before challenging material.
- • Depth-based risk previews
- • Required consent acknowledgment
- • Professional support recommendations
- • Gentle Mode softens challenging content
Mental Health Resources
Curated directory of professional support services accessible at any time.
- • Therapy finder tools
- • Support group directories
- • International crisis support
- • findahelpline.com for 200+ countries
Safe AI Design
Sage (Claude Opus) is built around non-diagnostic, non-directive Jungian principles.
- • Crisis detection runs before every response
- • Non-diagnostic approach enforced in system prompt
- • Tiered message limits (25 / 35 / 50 per day)
- • Fallback system if AI unavailable
Manipulation Detection
Silent protection against attempts to misuse the system. Transparent to legitimate users.
- • Exploitation pattern detection
- • Control language analysis
- • Educational redirection (not punishment)
- • Graduated escalation with professional referrals
User Control
Multiple depth levels so users can match the work to their current capacity.
- • 4 depth levels (Gentle to Intense)
- • Gentle Mode accessibility option
- • Pace at will, no forced progression
- • Reduced-motion support throughout
How the Safety System Works
Continuous Monitoring
Every message is analyzed in real-time for crisis indicators using keyword detection and pattern recognition before Sage formulates a response.
Risk Assessment
Concerning language triggers a 3-tier risk classification (low / medium / high) that determines which intervention path to take.
Immediate Intervention
High-risk detections bypass Sage entirely and display crisis resources directly: 988 Lifeline, Crisis Text Line, and SAMHSA Helpline.
Ongoing Access
Professional resources remain accessible throughout every session via the persistent crisis banner, not just after a trigger is detected.
Professional Standards & Limitations
What We Are
- A self-exploration and personal growth tool
- Grounded in established Jungian psychology
- Equipped with layered automated safety measures
- Designed to complement, not replace, professional therapy
- Transparent about our architecture and data practices
What We Are Not
- A replacement for professional therapy or psychiatry
- Able to diagnose mental health conditions
- Appropriate as the sole support during active crisis
- Supervised by licensed mental health professionals
- A substitute for human therapeutic relationships
Emergency Crisis Resources
United States
International
Questions About Our Safety Measures?
We're committed to transparency. If you have questions about our safety protocols or suggestions for improvement, we'd love to hear from you.