Can We Risk Waiting? Understanding Safety and Security in AI