An early-warning system that models disasters as instability transitions rather than isolated events. It combines force-based instability modeling with an interpretable ML escalation-risk layer to detect when hazards become disasters due to exposure growth, response delays, and buffer collapse.

🛡️ Protect AI outputs with JoySafety, an open-source framework ensuring content security and compliance for enterprise applications.