<p>This study explores the added impact of news reports about listed companies’ application of and focus on artificial intelligence (AI) on corporate default risk prediction under a deep learning framework. We designed a corporate default risk forecasting model for listed companies using four indicator groups: solvency and profitability, operational efficiency and growth, external macroeconomic conditions, and AI-related news. The predictive effectiveness of these indicator groups was evaluated using the Diebold-Mariano (DM) test. The results showed that deep learning overall outperformed convolutional neural networks (CNNs) and linear regression models, while matching random forest and XGBoost models. Solvency and profitability contributed the most to the improvement in predictive performance, followed by operational efficiency and growth, while AI news and external macroeconomic conditions were relatively weaker. Furthermore, the study finds that incorporating AI-related news into the model as incremental information significantly reduces prediction errors, even when all traditional financial and macroeconomic variables are included. This research demonstrates that AI-related news texts can provide a critical dynamic supplement to default risk prediction, confirming the potential value of AI news to corporate credit risk forecasting. It also offers a new theoretical perspective and a technical pathway for improving predictive model development and credit risk management practices.</p>