Automated Adversarial Stress Testing Using Generative Adversarial Networks for Dynamic Risk Scenario Synthesis

Traditional stress testing frameworks rely heavily on manually designed macroeconomic scenarios that often fail to capture the nonlinear and adversarial characteristics of modern financial systems. This study proposes an automated adversarial stress testing framework based on Generative Adversarial Networks (GANs) to generate realistic yet extreme risk scenarios. The framework integrates scenario generation, adversarial perturbation, and risk propagation modeling to evaluate systemic resilience under previously unseen market conditions. The proposed approach allows financial institutions to move beyond static stress scenarios by synthesizing data-driven stress environments that reflect complex market dependencies. Experimental evaluation using historical financial indicators demonstrates that the GAN-based system produces high-impact stress trajectories capable of revealing latent vulnerabilities in risk portfolios. The results suggest that adversarial generative models can significantly enhance scenario diversity, improve tail-risk discovery, and support proactive regulatory stress testing.