Authors: Panchal Aditya Krishnakant, Prof. J. M. Patade, Prof. M. C. Kolse, Prof. S. K. Said

Abstract: Cyber threats and hacker attacks are increasing rapidly with the growth of internet use, online transactions, cloud computing and digital communication, and traditional security systems that depend on predefined rules and signatures often fail against advanced attacks. Artificial Intelligence offers a way to identify suspicious activities, detect unusual patterns and prevent attacks. This paper reviews AI for detecting and preventing hacker attacks across the network, endpoint and user layers. It examines machine learning for network intrusion detection, deep learning for malware and ransomware detection from executable files and file-system behaviour, anomaly detection on system logs, phishing detection and user behaviour analytics for unauthorized access. The effectiveness and limitations of AI in security operations are discussed, including evaluation pitfalls, false positives and adversarial attacks in which attackers manipulate inputs to evade detectors. Mapping of detections to attacker tactics and automated response are also covered. A five-layer AI cyber-defence architecture comprising data collection, AI detection engines, correlation and threat intelligence, automated response and human analysts is proposed, along with research gaps, objectives, a comparison of methods and challenges. The review concludes that AI must be integrated with traditional controls, human expertise and sound policy.

DOI: https://doi.org/10.5281/zenodo.24219291