Machine Readership and Financial Reporting

In the age of Artificial Intelligence, most analysis on corporate financial statements are conducted by machine. We explore the disciplinary effect of such technology on firms’ financial misstatements. Firms respond to the increased machine readership by reducing financial misreporting that are sensitive to machine detection. As a falsification test, we find no evidence that machine readership is correlated to misreporting patterns that can be detected by traditional methods. This disciplinary effect is stronger among firms with more complex financial statements, more pressure from investors, and higher costs of earnings management.