The widespread adoption of the Android operating system makes its security critically important. While malware detection systems employ advanced techniques like static and dynamic code analysis along with antivirus software, they struggle against evolving obfuscation methods used by malicious developers. This work studies existing obfuscation techniques and proposes AndroGAN, a novel obfuscation as a security-through-obscurity approach using Generative Adversarial Networks (GANs). AndroGAN incorporates unique attributes including permissions, services, receivers, system calls, and sensitive tasks to enhance obfuscation. By training malware detectors and GAN-based generators on a newly curated dataset, AndroGAN simulates advanced evasion tactics, aiming to improve the robustness of detection systems. Obtained results demonstrate the abilities of AndroGAN in bypassing existing detectors, which contributes to strengthening Android security by addressing current limitations and advancing resilient malware detection methods. To promote transparency, collaboration, and further research, we have made both the curated dataset and the source code of AndroGAN freely available  here .