AI-Driven Cyber Risks and Financial Market Stability: Evidence from Volatility Markets and Risk Spillovers

The paper aims to investigate the effects of artificial intelligence (AI)-related cyber risks on financial market volatility, systemic spillovers, and forecasting performance for the banking sector. The study used daily stock return data for prominent Indian banking institutions over the period 2014-2025. A cyber risk index is constructed based on Google Trends data for cyber-related search intensity. The financial market volatility is analyzed using a GARCH model that incorporates the cyber risk index into the conditional variance equation. The systemic spillovers between financial institutions are examined using a vector autoregression (VAR) model. Additionally, the forecasting performance is also assessed using a Long Short-Term Memory (LSTM) neural network. The results from the empirical analysis indicate that cyber risks do indeed increase conditional market volatility. The results from the spillover analysis indicate that there is a high level of systematic interconnectedness between financial institutions. This is shown by a high value for the spillover index, which is approximately 89.98%. The results from the forecast evaluation indicate that the LSTM model does indeed outperform the traditional GARCH model. This is shown by lower values for the RMSE and MASE metrics, as well as statistically significant Diebold-Mariano test statistics -9.692. This reveals the increasing importance that cyber risks are having on financial instability. It also shows that the integration of AI-based forecasting techniques into traditional financial models does help in the evaluation of cyber-related financial risks. JEL Codes: G10, G21, C58, G28, G32