<p>Industry 4.0 technologies are proliferating, integrating Cyber-Physical Systems (CPS), the Industrial Internet of Things (IIoT), and cloud-connected industrial control systems, significantly expanding the attack surface for cyber threats. Consequently, traditional intrusion detection systems (IDS) are often insufficient in the face of the scale, complexity, and real-time nature of modern industrial environments.&#xa0;The objective of this research is to create and assess an integrated artificial intelligence (AI) system designed for immediate threat identification within Industry 4.0 environments. This innovative approach combines random forest (RF) algorithms, long short-term memory (LSTM) networks, and autoencoder techniques to effectively identify both recognized and emerging security risks across diverse manufacturing operations.&#xa0;Model training and evaluation were conducted using the TON IoT data, comprising telemetry, network, and log measurements from a smart factory simulation. Data preprocessing was done in terms of normalization, feature engineering, and class rebalancing using SMOTE. Models were evaluated in terms of accuracy, precision, recall, F1-score, and AUC, and cross-validation was used to assess robustness. Comparative benchmarking was used to compare state-of-the-art methods.&#xa0;The proposed framework achieved 95% accuracy and an average F1 Score of 95.0, which is much higher than that of existing models for detecting DoS, DDoS, Ransomware, and Backdoor attacks. It was shown to be highly flexible across three industrial settings (smart factory, power grid, and IIoT lab) and more robust and less training-intensive than state-of-the-art deep learning models.&#xa0;The suggested hybrid IDS framework is effective, readable, and scalable, and thus highly applicable for deployment in a real-time Industry 4.0 setting. Lastly, edge deployment, federated learning, and privacy-preserving AI are future directions for decentralized architecture.</p>