React OCRS: An AI-Driven Anonymous Online Reporting System Using Synergized Reasoning and Acting in Language Models

The rapid increase in cybercrime incidents worldwide has created an urgent need for efficient, secure, and accessible reporting systems. However, many cybercrime victims hesitate to report incidents due to privacy concerns, fear of identity exposure, and the complexity of traditional text-based reporting portals. This paper proposes ReACT_OCRS, an AI-driven anonymous online cybercrime reporting system that enables users to submit complaints through voice input without revealing their identity. The system integrates OpenAI Whisper for multilingual speech-to-text conversion, a hybrid LSTM and BERT model for intelligent complaint classification as genuine or false, and Federated Learning for privacy-preserving model updates. The backend is built using Java Spring Boot, providing a secure and scalable web application. Users are assigned unique anonymous IDs during registration, ensuring complete privacy. Encryption techniques protect all stored complaint data. Experimental results demonstrate that the proposed system achieves 91.4% complaint classification accuracy while significantly improving accessibility, efficiency, and user trust compared to conventional reporting methods. ReACT_OCRS represents a significant advancement in AI-powered E-Governance solutions for cybercrime reporting.