By taking advantage of security flaws to allow the execution of arbitrary instructions on a host system, command injection attacks represent a serious danger to the security of networked systems and web applications. Traditional detection mechanisms often fail to provide real-time detection and response capabilities, primarily due to their reliance on rule-based approaches or signature matching, which are limited by their inability to adapt to evolving attack patterns. The convolutional neural network (CNN) model controlled by AI is presented in this paper as a real-time command injection attack detection tool. The proposed model leverages deep learning techniques to automatically learn complex patterns from vast datasets of network traffic and application logs, allowing it to identify both known and unknown attack vectors with high accuracy. The model is designed to process incoming data streams in real time, applying convolutional layers to extract hierarchical features that represent potential attack signatures. By employing advanced training methodologies and optimization strategies, the CNN model is able to distinguish between normal and malicious commands effectively, reducing false positives while maintaining rapid detection rates.