The Logic Matrix Safety System
Every piece of content that reaches teachers passes through our comprehensive safety validation system. Nothing bypasses these checks—this is our liability shield and what keeps Taughtful defensible.
The Single Entry Point Rule
"Nothing returns to teachers without passing through the Logic Matrix."
This architectural principle ensures that all content—whether from Bloomie (Access Designer), Fin (Friction Notes), Ledger (Teaching Memory), or Evidence Export—undergoes the same rigorous safety validation before teachers see it. This stops inconsistent logic across tools and ensures uniform safety standards.
Four-Layer Safety Validation
All content must pass all four safety categories before reaching teachers. If any check fails, content is blocked or sanitized.
Clinical Language Detection
Prevents diagnostic, therapeutic, or medical terminology from appearing in teacher-facing content. Ensures all language remains educational and non-clinical.
Checks Performed:
- Scans for banned clinical terms (diagnosis, therapy, treatment, disorder, etc.)
- Detects diagnostic language patterns
- Identifies therapeutic claims
- Flags medical advice terminology
Diagnostic Claim Blocking
Blocks any content that suggests diagnosis, clinical assessment, or identification of conditions. Teachers see observable behaviors, not diagnostic labels.
Checks Performed:
- Detects diagnostic claim patterns ('has ADHD', 'shows signs of', etc.)
- Blocks content suggesting clinical assessment
- Prevents condition identification language
- Flags diagnostic inference statements
Restrictive Practice Guard
Blocks any suggestions of coercive, restrictive, or punitive practices. Ensures all strategies are supportive and trauma-informed.
Checks Performed:
- Detects restrictive practice terms (timeout, isolation, restraint, etc.)
- Flags coercive language ('forced to', 'must comply')
- Blocks punitive suggestions
- Prevents exclusionary practices
Tier Creep Prevention
Ensures strategies stay within authorized tier levels (Tier 0-1 by default). Prevents escalation to specialist or clinical interventions.
Checks Performed:
- Validates tier indicators match authorized level
- Detects tier escalation language
- Flags specialist intervention suggestions
- Ensures classroom-level appropriateness
How Content Gets Processed
Every piece of content follows this validation pipeline before reaching teachers.
Content Input
Content enters the Logic Matrix from any source (Bloomie, Fin, Ledger, Evidence Export)
Safety Checks Run
All four safety categories are checked simultaneously: Clinical Language, Diagnostic Claims, Restrictive Practices, Tier Creep
Violation Assessment
System determines severity: Block (content rejected), Warn (flagged for review), or Sanitize (terms replaced with safe alternatives)
Content Sanitization
If sanitizable violations found, clinical terms are replaced with teacher-friendly alternatives (e.g., 'diagnosis' → 'educational observation')
Safe Content Delivery
Only validated, safe content reaches teachers. All violations are logged for audit but never shown to teachers
Banned Terms & Patterns
These terms are automatically detected and blocked or sanitized to ensure teacher-facing content remains educational and non-clinical.
Clinical Terms
SanitizedDiagnostic Labels
BlockedRestrictive Practices
BlockedMedical/Professional
SanitizedSafe Language Translation
When clinical terms are detected, they're automatically replaced with teacher-friendly alternatives that maintain educational focus.
Complete Audit Trail
Every safety check is logged with full details for compliance and accountability:
Important: Violations are logged for internal audit but never shown to teachers. Teachers only see safe, validated content.