An Innovative ESG Score Prediction and a Greenwashing Detection Model Using Financial Data Based on Artificial Neural Networks

ABSTRACT Rating agencies exhibit significant differences in their company ESG evaluation methodologies, due to a lack of standardization despite the importance of the process. Recently, AI models for ESG performance prediction using companies' financial data enabled a low‐cost, feasible, and highly accurate evaluation of companies' sustainable strategies. In this paper, a new ESG score prediction model using financial data is proposed, which is based on Artificial Neural Networks and the Levenberg–Marquard training algorithm. The proposed model is evaluated on companies listed in the S&P 500, using two independent datasets provided by two rating agencies, and it outperforms previous work in predicting ESG scores. Financial data and reported ESG score evaluations are incorporated into the training procedure, allowing the neural network to learn the hidden rules between financial data and ESG performance. A novel Artificial Neural Network Greenwashing detection model is proposed for the first time. This model detects potential greenwashing by quantifying discrepancies between predicted and reported ESG performance, offering a strategic tool to detect misleading claims and strengthen ESG accountability.