Emergency communication channels are increasingly flooded with messages that are machine-generated, outdated, or deliberately misleading, which delays response and erodes public trust. Existing detectors usually depend on a single signal, such as writing style or propagation pattern, and rarely explain why a message is flagged. This paper proposes Adaptive Emergency Message Reliability Assessment (AEMRA), a framework that fuses five signals: AI-generation likelihood, source credibility, spatial consistency, temporal consistency, and retrieved evidence support. A context-aware gating network assigns signal weights and renormalizes over the signals that are actually available, so the method keeps working when a geotag or source profile is missing. AEMRA outputs a reliability score, a risk-priority rank for moderators, and a per-signal explanation. We evaluate AEMRA on [datasets] against [baselines] using [metrics]. Results show [main finding], and ablation shows [finding on adaptive weighting and missing signals].

Maseeha Bhanu, Amulya D R, G V Abishek Bhargav, Ashwini C J, Suhas S L, "A Reliability-Aware Framework for Detecting AI-Generated and Misleading Emergency Messages Using Multi-Signal Evidence Analysis" International Journal of Scientific Research and Engineering Development, V9(6): Page(115-118) November - December 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.