N°26-25: Artificial Intelligence and Risk-Taking in Banking

We study how banks’ adoption of artificial intelligence (AI) affects risk-taking and stability using data on major listed Chinese banks (2013–2024). We construct an AI-intensity index from agent adoption, cloud computing, and big data measures. Higher AI intensity is associated with greater risk and lower stability: a one–standard deviation increase reduces the Z‑score by 0.25 units (about 22%) next year. This operates mainly via higher earnings volatility and lower profitability, not weaker capitalization. The effect is stronger for banks with low liquidity, unstable funding, and weak loss-absorption, implying AI can amplify risk and requires tighter prudential oversight.