Sentiment, Risk, and Market Shock: The Impact of AI Risk Disclosures on Stock Reactions to DeepSeek_R1

12 Pages Posted: 6 May 2025

See all articles by Yue Yu

Yue Yu

University of Wollongong, Wollongong NSW Australia

Bin Liu

University of Wollongong

Tina Prodromou

The University of Wollongong

Sandy Suardi

University of Wollongong

Date Written: April 15, 2025

Abstract

This study explores how the sentiment of AI risk disclosures affects stock market reactions to the DeepSeek_R1 event. Analysing 10-K reports of IT companies using FinBERT, we find that firms with more optimistic AI risk disclosures experience significantly stronger negative market reactions. This suggests that overly optimistic disclosures may lead investors to underestimate potential AI risks. Our research offers new empirical evidence on the downsides of optimistic AI risk disclosures and underscores the need for balanced risk communication. The findings have practical implications for investors and provide insights for managers and regulators addressing emerging technological risk disclosure issues.

Keywords: Risk disclosure, Sentiment, DeepSeek, Market reaction, AI risks

Suggested Citation

Yu, Yue and Liu, Bin and Prodromou, Tina and Suardi, Sandy, Sentiment, Risk, and Market Shock: The Impact of AI Risk Disclosures on Stock Reactions to DeepSeek_R1 (April 15, 2025). Available at SSRN: https://ssrn.com/abstract=5218946 or http://dx.doi.org/10.2139/ssrn.5218946

Yue Yu

University of Wollongong, Wollongong NSW Australia ( email )

Bin Liu (Contact Author)

University of Wollongong ( email )

Northfields Avenue
Wollongong, New South Wales 2522
Australia

Tina Prodromou

The University of Wollongong ( email )

Northfields Avenue
Wollongong, NSW 2522
Australia

HOME PAGE: http://https://scholars.uow.edu.au/display/tina_prodromou

Sandy Suardi

University of Wollongong ( email )

Northfields Avenue
Wollongong, New South Wales 2522
Australia

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
15
Abstract Views
38
PlumX Metrics