Artificial Intelligence Models for Detecting Greenwashing in UK ESG and Green Finance Projects

This study examines the application of artificial intelligence models for detecting greenwashing practices in UK Environmental, Social, and Governance projects and green finance initiatives. The research addresses the growing concern over misleading sustainability claims in light of the UK Financial Conduct Authority's anti-greenwashing rule implemented in May 2024. Employing a mixed-methods approach, this study develops a comprehensive framework integrating Natural Language Processing techniques, specifically transformer-based models including BERT and ClimateBERT, with machine learning algorithms such as XGBoost and Random Forest for quantitative prediction and classification. The methodology incorporates a dataset of UK-based companies' sustainability reports, ESG disclosures, and green finance documentation from 2018 to 2024, comprising 487 firms across multiple sectors. The quantitative analysis utilizes a dual approach: textual analysis through NLP models achieving 86.34% accuracy in identifying greenwashing risk patterns, and financial-ESG divergence analysis using optimized machine learning models with R² values of 0.9790. Key findings reveal that AI models can effectively identify discrepancies between ESG disclosure scores and actual environmental performance, with firm size, governance structure, and financial constraints emerging as significant predictors of greenwashing behaviour. The study contributes to the literature by providing a robust, scalable methodology for regulatory bodies and investors to enhance transparency in sustainable finance markets, ultimately supporting the UK's commitment to achieving net-zero emissions targets.