Media
2 items
Description
Family safety for AI sites: local SCAB (0–18) & PRIS (0–100) scores, kid/parent modes, on‑device policy checks. SafeSpaceAI Family – SCAB/PRIS Protection for Children, Parents, and Schools SafeSpaceAI Family is a privacy-first Chrome extension designed to help families, educators, and schools create safe, transparent, and age-appropriate AI experiences. It runs entirely on-device by default, evaluating visible or selected text on AI chat websites with two complementary scores: • SCAB (0–18): Six behavioral domains indicating ethical and safety risk. • PRIS (0–100): A psychosis-risk interaction score that flags concerning conversational cues across six domains. The extension provides real-time overlays, warnings, and optional content blurring when AI conversations exceed family-defined thresholds. Parents and teachers can adjust settings, review safety interventions, and export redacted data for audit or reporting. ⸻ Key Features 👨👩👧 Kid Mode & Parent/Teacher Mode • Kid Mode (default): Simple, safe experience with automated scoring and protective overlays. • Parent/Teacher Mode: Unlocks dashboards, thresholds, and export features. Access is PIN-gated for security. 🔍 Real-Time Scoring • Every AI message is checked with SCAB and PRIS heuristics. • Risk levels are color-coded: ✅ Safe, ⚠️ Warn, ⛔ Block. • Optional blurring prevents exposure to unsafe responses. 📊 Transparent Metrics • SCAB (0–18) scored across six domains: Sovereignty, Coherence, Alignment, Boundaries, Epistemics, and Governance. • PRIS (0–100) scored across six cues: Content risk, Suggestibility, Isolation/Paranoia, Sleep/Mania cues, Reality testing, and Urgency/Self-harm. • Thresholds can be customized to reflect family, classroom, or school policies. 📚 Child-Friendly Explanations & Lessons • When content is blocked, a “Why Blocked” view explains in plain, child-appropriate language. • Optional literacy modules teach safe AI use and digital citizenship. 🛡️ Privacy by Design • Offline by default: All analysis happens locally, inside the browser. • No remote code: The extension never loads external scripts or libraries. • No data collection: No personal data, browsing history, or conversation content is sent to any server. • Optional Hybrid Mode: For school districts, admins may configure a relay server. Content is locally redacted before transmission, and fallback to offline occurs automatically if the relay is unavailable. 📋 Parental & Educator Controls • Set thresholds for warnings and blocks. • PIN-gated parent/teacher mode to prevent tampering. • Export JSON settings and evidence bundles. • Enable per-site protection only when approved by the parent or teacher. 🏫 Classroom & School Use • Designed for families and classrooms. • Supports district-level presets and policies. • Helps teachers supervise AI usage without exposing raw content. • MDM-friendly design for schools. ⸻ How It Works 1. Enable on a site: With ChatGPT, Gemini, Claude, Perplexity, or another AI site open, click Enable on this site to activate protection. 2. Local analysis: Messages are scanned by lightweight heuristic models (SCAB/PRIS). 3. Overlay & action: A badge shows the scores. If thresholds are exceeded, the message may be blurred and a warning shown. 4. Parent oversight: Parents/teachers can unlock dashboards to adjust thresholds, export evidence, or toggle hybrid relay settings. ⸻ Benefits • Safe: Children and students are protected from unsafe or manipulative AI outputs. • Private: No personal data leaves the device; hybrid relay is optional and district-controlled. • Flexible: Works across multiple AI sites; parents and teachers can tune thresholds. • Educational: Helps children learn safe AI practices with clear feedback and optional lessons. • Trusted: Built with privacy-by-design principles and aligned with child safety standards (COPPA/AADC). ⸻ Permissions Explained • Storage: Saves thresholds, modes, and PIN locally. • ActiveTab: Needed to enable protection on the site you choose. • Scripting: Injects overlay and scoring logic only when you enable it. • Optional Host Permissions: Requested only when you enable protection on a specific AI site. No unnecessary permissions are used, and no background data collection occurs. ⸻ Disclaimer SafeSpaceAI Family is a safety and educational tool. It provides heuristic scoring and warnings but does not replace parental guidance, teacher supervision, or professional mental health support. It should not be used for medical diagnosis or emergency intervention. If you encounter content related to self-harm, please contact local crisis resources or emergency services immediately.