Applied Artificial Intelligence Solutions for Accessible European Emergency Services

Artificial intelligence holds real promises for making emergency communication more accessible, yet in the European Union, its actual integration into 112 systems remains patchy and, in most cases, still experimental. This study does not aim to add to the theoretical literature on inclusive technologies; instead, it asks what an applied AI-enabled accessibility framework might actually look like in practice, within nextgeneration emergency communication systems. Drawing on a structured qualitative review of recent technical benchmarks and policy documents, we propose a Technical Reference Architecture aligned with current European regulatory requirements. The central question we examine is how far AI can genuinely reduce communication barriers for people with disabilities, and perhaps more importantly, what is still stopping this from happening on a large scale. Our findings point to the value of an AI orchestration layer that brings together automatic speech recognition (ASR), Natural Language Processing (NLP), and explainable artificial intelligence (XAI), particularly for callers with sensory or cognitive impairments. The architecture we propose is not built around predictive accuracy alone. It is designed to keep human dispatchers in the loop, making AI-assisted triage decisions traceable and open to challenges. The study also attempts to translate the requirements of the EU Artificial Intelligence Act (2024) into something more concrete: a structured engineering framework that connects system design with the governance principles of transparency and explainability. The broader conclusion is easy to state, even if achieving it is not. Accessibility in emergency communication should be treated as a strategic capability, not an afterthought. No citizens should be left behind during a crisis.