AI Disclosure, Signal Quality and Washing: Evidence From Chinese Listed Firms

ABSTRACT Using Chinese A‐share listed firms from 2016 to 2024, we examine whether Firm‐Specific AI Disclosure predicts subsequent observable AI activity and when disclosure becomes disconnected from action. AI disclosure is most consistently associated with future patenting, while evidence for hiring and broader capital investment is shorter‐lived, supporting an informative‐but‐imperfect signal interpretation. We define AI washing as multidimensional disclosure–action mismatch and find that greater mismatch is associated with AI‐specific stock‐exchange scrutiny. However, inquiry is not followed by broad and persistent realignment: fewer than half of classifiable responses meet our Substantive‐Evidence threshold, consistent with a Detection–Correction Gap.