Do AI outages disrupt financial markets? Evidence from OpenAI

Document Type

Article

Publication Title

Finance Research Letters

Publication Date

8-1-2026

Abstract

This study examines how financial markets respond to disruptions in AI-based information processing. We examine 207 documented OpenAI service outages between April 2021 and August 2025 as quasi-exogenous shocks to market information infrastructure. Across 11,978 affected firms, abnormal trading turnover increases on outage days. Liquidity drops, with Amihud's illiquidity measure implying a 580 basis-point (bps) higher price impact per $1M of trading volume during outages. Abnormal absolute returns are 1.4 bps lower on outage days and abnormal signed returns are 9.2 bps lower. The impact of outage duration is non-monotonic: returns are most affected by short outages, turnover by intermediate-length outages, and illiquidity by long outages. Overall, our findings suggest that AI systems are now an integral component of modern market infrastructure and that AI outages are associated with significant disruptions in trading activity.

Volume

104

DOI

10.1016/j.frl.2026.110162

ISSN

15446123

E-ISSN

15446131

Link to Full Text

Share

 
COinS