Pharoah Suite/aegis
Aegis — Age Assurance & AI-Safety Guardrail
⚠ Age assurance is simulated — no vendor is connected and no real age check is performed
AI-safety guardrail that classifies chat messages against child-safety policy and escalates or redirects harm, with age assurance simulated rather than live. Live-probed 2026-08-18: 9/9 harm prompts caught across English, Spanish, French, German and Portuguese including deliberately misspelled ones, 4/4 ordinary messages passed untouched, and languages outside those five refused rather than let through. No age-assurance vendor is connected, so no real age check is performed. Evaluation build, not a compliance guarantee: extremism, eating-disorder, drug-synthesis and gambling content are not detected today, and messages are judged one at a time with no conversation history.
Not for sale yet · evaluation access only