Malware detection systems that incorporate machine learning algorithms have received considerable attention because of their impressive ability to identify and combat emerging and sophisticated malware threats effectively. These systems rely on advanced techniques to analyse vast amounts of data and recognise patterns that distinguish between harmless and malicious software. The development of Generative Adversarial Networks [1] has enabled innovative techniques to enhance the capability of the malware detection systems, Mal-GAN [2]. In this research paper, an innovative approach called Mal-cGAN is introduced, aiming to enhance the performance of malware detection systems that solely work on image representations of software by synthesising image representations of malware. Drawing inspiration from the architecture of Conditional Generative Adversarial Networks (cGANs) [3], the method employs a “U-Net” [4] based generator that facilitates seamless communication between the Malware Detection System and Substitute detector with the Substitute Detector performing the role of a discriminator from typical GANs. This effective collaboration enables the generator to produce practical and highly realistic malware image representations, which in turn can bypass the detection systems and eventually used to enhance the performance of the Malware Detection System. Thorough evaluations of the model demonstrate that the Mal-cGAN generator successfully synthesizes accurate and informative image representations of malware. Notably, training the Malware Detection System on these synthesized samples leads to a substantial increase in accuracy, improving from 83.47% to 93.61%. The results depicting synthesised image representations of malware and the classification metrics highlight the effectiveness of the proposed approach and its potential to revolutionise the field of malware detection using machine learning.