The rapid expansion of digital communication platforms has significantly improved the speed at which emergency information reaches the public; however, it has also increased the circulation of misleading alerts and fabricated warnings that can disrupt coordinated response operations and undermine public trust. Traditional verification approaches, typically based on manual validation or static rule-based filtering, struggle to handle the scale and variability of contemporary communication streams. To address these limitations, this study presents an Artificial Intelligence (AI)-driven cloud-based framework for detecting fake emergency alerts and misinformation using advanced Natural Language Processing (NLP) and machine learning techniques. The proposed system integrates scalable cloud infrastructure with supervised classification models to enable continuous monitoring and real-time analysis of high-volume textual data. Feature engineering procedures, including semantic vector embeddings, contextual similarity scoring, and linguistic pattern extraction, were applied to enhance model discrimination capability. The framework was evaluated using benchmark crisis communication and social media datasets comprising over 50,000 labeled emergency-related messages. Experimental results indicate that the proposed model achieved an overall classification accuracy of 96.4%, with a precision of 95.8%, recall of 96.9%, and F1-score of 96.3%. The receiver operating characteristic analysis yielded an area under the curve (ROC-AUC) value of 0.982, demonstrating strong separability between legitimate and fraudulent alerts. In addition, the system maintained a low false positive rate of 2.7% while processing large message streams. Performance analysis further revealed that the cloud-based architecture sustained stable response times, with an average message processing latency of 1.8 seconds under normal workload conditions and 2.6 seconds during peak traffic scenarios involving up to 50,000 concurrent messages. Scalability testing confirmed that the system maintained consistent detection accuracy above 95% as dataset size increased, indicating reliable performance in large-scale operational settings. These findings demonstrate that the proposed AI-driven framework provides an effective, scalable, and reliable mechanism for identifying misinformation in emergency communication networks, thereby supporting timely decision-making and strengthening the resilience of public safety information systems.

Keywords: Fake emergency alerts, misinformation detection, cloud computing, natural language processing, machine learning, public safety systems, real-time monitoring