An AI-Powered Cybersecurity and Digital Safety Companion for Real-Time Threat Detection and User Protection
Abstract
The rapid proliferation of cyber threats including phishing attacks, deepfake-based abuse, non-consensual intimate image (NCII) distribution, and physical safety emergencies has created an urgent demand for a unified, intelligent protection platform. Existing solutions operate in isolation: URL-based phishing detectors lack visual context, deepfake research rarely delivers prevention tools, and SOS apps offer no cyber integration. Proposed system SAFO (Smart AI Framework for Online Safety), an AI-powered cyber security and digital safety companion app that consolidates multiple defenses into a single lightweight Android platform. SAFO integrates an ensemble phishing detection engine combining URL analysis, OCR, and Visual Language Model (VLM) logo verification; an AI-driven deepfake and NCII media forensics module; a voice-and-camera-activated SOS emergency system triggered by the phrase "HELP ME"; an OSINT-based Reputation Scanner for digital identity monitoring; a Crowd-Sourced Threat Map for geo-fenced community alerts; a Criminal Legal Draft Generator; and an AI Chatbot for real-time cyber security guidance. Privacy-first model updates are achieved through federated learning, ensuring no raw user data leaves the device. Evaluation demonstrates phishing detection accuracy of 85%, deepfake detection accuracy of 90%, and SOS trigger precision of 95%, outperforming fragmented existing systems across all benchmarks. SAFO offers a holistic, mobile-optimized, and explainable security ecosystem for everyday users.