Kumar, Venkataramani2025-09-192025-09-192025-07-28https://hdl.handle.net/20.500.11801/7473This was a presentation created to be presented at NAECON 2025 on security threats on edge AI. However, it wasn't presented there. But this presentation is included for reference or additional reading.With the evolution of advanced next-generation applications the need to meet the low-latency requirements is necessary. By processing the information at edge, it is feasible to expedite the response. However, the edge processing presents cybersecurity threats. Most of the existing works focus on resolving one aspect such as anomaly detection, or adaptive learning. However, through our work we propose an edge-enabled intelligent framework that offers three main functionalities namely threat level classification, response code prediction, and anomaly score prediction. The proposed monolithic framework is validated on open-source dataset with numerous features. Preliminary evaluation results highlight the significance of our framework.en-USAttribution-NonCommercial 4.0 Internationalhttp://creativecommons.org/licenses/by-nc/4.0/Edge AIsecurity threatsCyber intelligence (Computer security)CodingPrediction (Logic)Presentation(c) 2025 Venkataramani Kumar