[Phase 5.3] Safety Transparency Interface

[robinson]

Goal

Give parents and teachers clear, understandable visibility into the safety layer
that governs all AI interactions.

Scope

• "Safety settings" panel in the dashboard: active content filters, topic
  allow/block lists, age-appropriateness level.
• Per-session safety summary: how many messages were filtered, why.
• Link between flagged interaction and the specific safety rule that triggered it.
• Parent-configurable topic blocks (e.g., block "social media advice").

Acceptance Criteria

• Safety settings panel lists all active rules in plain language (no jargon).
• Each flagged interaction links to the rule that triggered the flag.
• Parent-added topic blocks apply within 1 session restart.
• Audit log records all safety rule changes with timestamp and parent ID.

References

• design/oversight-dashboard/README.md — Safety Controls section
• docs/VOICE_CONSENT_SAFETY.md
• eule-core/src/storage/sqlite.rs

Phase

Phase 5 — Parental Dashboard Redesign · Week 3–5