Artificial Intelligence and Workplace Misconduct

This paper studies the impact of artificial intelligence (AI) adoption on workplace misconduct. We combine firm-level data on financials, corporate governance, and workplace misconduct with a measure of AI adoption based on employees' AI-related skills. To identify causal effects, we exploit the 2015 release of Google TensorFlow as a plausibly exogenous shock within a difference-in-differences framework. We find that firms with higher pre-treatment AI intensity experience significant and persistent declines in workplace violations and penalty amounts after 2015. The effects operate primarily through productivity-enhancing complementarities, while labor-adjustment channels play no role. Benefits are concentrated among larger, intangible-and organizational-capital-intensive firms, which highlights uneven gains from AI adoption.