Poster: EVERESTS: A Mobile-First, AI-Driven Cyber-Safety Companion for Older Adults
Authors: Apeh, E., Awad, A., Khamis, M.
Journal: Mobisys Companion 2026 Companion Proceedings of the 24th ACM Annual International Conference on Mobile Systems Applications and Services Part of Mobisys 2026
Publication Date: 20/06/2026
Pages: 51-52
DOI: 10.1145/3812835.3814875
Abstract:Older adults face escalating risks from rapidly evolving online and phone-based scams powered by AI-generated content and personalised social-engineering techniques. Existing cyber-safety tools often rely on static guidance, lack real-time adaptability, and rarely integrate with the mobile interaction environments where scams actively occur. We present EVERESTS, a mobile-first cyber-safety system that delivers personalised, context-aware scam-awareness guidance using a hybrid cloud/on-device architecture. EVERESTS continuously ingests live intelligence from OSINT, SOCMINT, and geo-linked sources, curates emerging threats, and generates accessible explanations and proactive alerts via a Retrieval Augmented Generation (RAG) workflow powered by a large language model (RAG-LLM). Our prototype includes two interaction pathways: Dahlia, a conversational assistant for text-based guidance, and Mark, a text-to-speech voice persona for phone-call-style proactive warnings. EVERESTS aims to support vulnerable users through trustworthy, personalised interactions while maintaining strong privacy and security boundaries. We describe the system's architecture, key design principles, and prototype implementation.
Source: Scopus