Using AI to assess the decision-usefulness of corporates’ nature-related disclosures

Abstract Companies’ nature-related disclosure is insufficient for decision making. This situation is unlikely to improve as long as it remains voluntary, even under initiatives such as the Task Force on Nature-related Financial Disclosures (TNFD). To investigate the decision-usefulness of these disclosures, we develop A sk N ature , an AI-powered Question Answering (QA) tool that uses Retrieval Augmented Generation (RAG) to analyze corporate reports and score their alignment with TNFD recommendations. We apply A sk N ature to the Corporate Sustainability Reports (CSRs) of the Nature Action 100 (NA100) companies, a group selected for their systemic importance in reversing nature loss. Our analysis reveals a disconnect between reporting and reality: disclosures are dominated by positive sentiment despite ongoing biodiversity loss, geographically ignore many of the world’s most critical natural assets, and contain commitments that score high on declaration but low on specificity and even lower on legal enforceability. Overall, current CSR disclosures, although aligned with the TNFD, are not sufficiently decision-useful for stakeholders and lack legal enforceability.