Investor AI Monitoring Capability and ESG Disclosure Granularity: Evidence from East African Banking

The rapid proliferation of artificial intelligence tools among institutional investors is reshaping corporate governance and accountability, yet its consequences for ESG disclosure quality in frontier markets remain poorly understood. This study examined one precisely bounded question: does investor AI monitoring capability cause East African commercial banks to disclose ESG information more granularly? The analysis was grounded in agency, signaling, and institutional theories, which together position AI capability as a governance mechanism that conditions the depth — not merely the breadth — of ESG reporting by altering the strategic cost of disclosure imprecision. Using hand-coded ESG disclosure data from 31 commercial banks — 23 domestic private and 8 globally affiliated — across Kenya, Tanzania, Uganda, Rwanda, and Ethiopia (2018–2024), and surveys of 418 institutional investors, the study constructs an ownership-weighted AI capability measure and a comprehensive ESG disclosure granularity index comprising 47 items across environmental, social, and governance dimensions. The empirical analysis employed OLS, instrumental variable estimation exploiting EU SFDR mandates, staggered difference-in-differences, and event studies, controlling for firm size, profitability, leverage, ownership structure, and board characteristics. Investor AI capability is a strong, robust predictor of disclosure granularity (β = 0.52, p < 0.001, ΔR² = 0.22), with a one-standard-deviation increase associating with a 14.9-point granularity rise. Results were consistent across all five identification strategies (β range: 0.49–0.68). Crucially, actual ESG performance does not predict granularity, and AI effects are strongest among poor ESG performers — consistent with strategic impression management rather than genuine accountability. The study introduced the concept of 'algorithmic greenwashing': the production of granular, machine-readable disclosures optimised for AI detection without substantive improvement to underlying ESG practices. Regulators should mandate granularity standards with independent verification mechanisms, and must not treat algorithmically optimised disclosure as a proxy for genuine sustainability progress. Investors and bank boards must ensure detailed reporting reflects substantive rather than reputational compliance.