AI strategy, earnings management, and corporate fraud: Evidence from listed firms in China

This study examines how heterogeneous artificial intelligence (AI) strategies affect corporate fraud, addressing a gap in the literature that has largely focused on AI's governance role from a technological perspective while overlooking firms' underlying adoption motivations. Using panel data on Chinese A-share listed firms from 2013 to 2023, we distinguish between symbolic and substantive AI strategies and analyze their differential effects on corporate fraud. The results show that symbolic AI adoption significantly increases fraud risk, particularly among highly financialized firms, non-manufacturing firms, and firms operating under high uncertainty, whereas substantive AI adoption has no direct effect on fraud incidence. Mechanism analysis reveals that symbolic AI increases fraud risk indirectly through accrual-based earnings management, suggesting that opportunistic financial reporting constitutes an important transmission mechanism. In addition, we find that non-standard audit opinions significantly weaken the positive association between symbolic AI adoption and corporate fraud, highlighting the disciplinary role of external audit oversight. Overall, these findings underscore the importance of organisational motivation in shaping the economic consequences of AI adoption and offer policy-relevant implications for fostering more rational and transparent use of emerging technologies.