AI rhetoric and audit prudence: a sociotechnical analysis of AI disclosure intensity in China

Corporate AI rhetoric has become a powerful sociotechnical force. This study examines how AI rhetoric in corporate disclosures affects auditor behavior. We measure AI rhetoric as AI disclosure intensity via the frequency of AI keywords, a techno-discursive practice embedded within China’s governance networks. Based on STS and institutional theory, we contend that extensive AI rhetoric gives rise to interpretive ambiguity, leading auditors to adopt higher audit prudence. Audit delay serves as the primary indicator, with audit fees used for robustness tests. Using 31,389 firm-year observations of Chinese A-share listed firms from 2016 to 2024, this study finds a significantly positive correlation between AI rhetoric and audit prudence. Investment scale acts as the primary mediator, while total factor productivity and analyst attention play secondary mediating roles. Further contextual analysis shows that post-2023 AI regulations reduce interpretive ambiguity and weaken this relationship; Big 4 audits have a similar moderating effect, whereas higher profitability strengthens it. Neither state-owned enterprise status nor AI industry affiliation generates significant moderating effects, which challenges the conventional view that firm identity brings institutional legitimacy. This paper identifies a “governance paradox of AI rhetoric”: firms attempt to signal innovation via AI rhetoric but trigger heightened institutional caution. The findings provide references for regulators, auditors and enterprises to balance innovation signaling and accountable AI governance.