Spotlight Safety Demo
Interactive chat moderation powered by Spotlight. Every message is evaluated in real-time with per-user multi-turn context tracking.
connecting…
Ready
Welcome to Spotlight Chat Moderation (v3.2)!
Send messages below to test real-time safety classification powered by
💡 Features to Test: Click
Send messages below to test real-time safety classification powered by
student_6L_v3_2. Messages are evaluated across 95 labels (harms, conversational framing, stance, and speech acts) with multi-turn context tracking and policy exception gates.💡 Features to Test: Click
⚠️ "go" then "die" for multi-turn tracking, 🧟 "kill the zombies" for NPC combat exception, or switch Surface to test short-text shouting suppression!
Safety Inspector
Select a message
🔍
Send or click any chat message to inspect its safety verdict, label probabilities, and multi-turn debug signals.