A Cyber Kavach: GENAI-Powered Multi-Model Cyber Threat & Legal Aid System

Cyber Kavach is a multi-modal Cyber Threat Detection and Legal Aid System, utilizing generative AI (GenAI) capabilities and built for the current increase in the number of people using (and misusing) the Internet; particularly, in India where the Internet is growing rapidly and infamously diversified. The rise of increasingly sophisticated cyber threats like phishing, financial fraud, impersonating attacks, and malicious/harmful links that can occur online. However, existing detection systems do not provide actionable guidance based upon contextual intent for vulnerable digital users. This proposal describes an advanced intelligent, full-stack web application designed for this purpose. The application will use Large Language Models (LLMs) like HuggingFace’s Mistral7B to analyze all suspicious input (input that may be associated with known malicious activity) including text messages, URL’s, images, and documents – without requiring users to have any technical knowledge. In addition, the system will utilize an 11-point, custom-built risk scoring system that evaluates characteristics of a threat (urgency of the associated language, suspiciousness of the associated domain, request for OTP etc.; impersonation patterns); and will generate two types of outputs for each cyber threat detected: 1)an interpretable risk score to represent the level of risk associated with the threat and a clear description of the threat in plain English. Unlike existing threat-detection systems, Cyber Kavach includes a unique legal assistance module that will automatically map detected threats to provisions of the Information Technology Act, 2000 and the Indian Penal Code, and generate structured, ready-for-filing FIR drafts in at least 3 languages (English, Hindi, and Telugu) to help eliminate a significant barrier experienced by victims when attempting to report cybercrime. The Cyber Kavach Platform employs a three-tier architecture with a React frontend, a FastAPI backend, and secure AI integrations while protecting user data by performing all AI interactions on the server, providing users with a simple and expert user interface. The platform has been tested with real-life scam situations and has been able to identify threats accurately and provide timely responses, thus proving to be an effective and scalable method of improving cyber safety. By providing real-time threat detection, multilingual capabilities, and automated legal assistance, Cyber Kavach allows users to successfully identify and respond to cyber threats, which helps build a more informed and resilient digital community