Authors: Sandhya R. Bista
Abstract: Generative Artificial Intelligence (AI) has rapidly become a transformative yet paradoxical force in the domain of cybersecurity. Its dual-edged nature capable of both fortifying defenses and amplifying cyber threats has redefined the way organizations approach malware detection, prevention, and response. Traditional cybersecurity models, which rely heavily on signature-based detection and heuristic methods, are increasingly inadequate against polymorphic, evasive, and zero-day malware variants. These conventional systems lack the adaptive capacity to counter attackers who continuously modify malicious code to escape static defense algorithms. In contrast, generative AI introduces a new paradigm in which defense systems evolve dynamically, learning from both real and simulated threats to anticipate and neutralize future attacks before they occur. Generative AI models such as Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and transformer-based architectures have been instrumental in driving this evolution. GANs, for instance, can simulate sophisticated attack patterns, enabling security systems to train against artificially generated malware samples that replicate real-world adversarial behavior. Similarly, transformer-based models enhance contextual awareness and anomaly detection by processing vast streams of network, behavioral, and endpoint data in real time. This fusion of generative modeling and adaptive learning fosters proactive defense strategies capable of identifying subtle deviations indicative of malicious intent long before damage is inflicted.
DOI: http://doi.org/10.5281/zenodo.17879335
