Large Language Model Application for Regulatory Horizon Scanning: Case Study on ESG Greenwashing Regulations

This white paper explores the application of Generative AI, specifically Large Language Models (LLMs), to enhance regulatory horizon scanning within financial services. Using the Financial Conduct Authority’s (FCA) 2024 anti-greenwashing rule as a case study, we demonstrate how LLMs can be integrated into the strategic foresight process to detect early regulatory signals, analyse stakeholder feedback, and forecast future regulatory developments.

Our framework builds upon the traditional horizon scanning model, comprising exploration, assessment, application, and continuation, and incorporates advanced text analysis techniques including semantic similarity testing with models such as BERT and RoBERTa.

The study shows that LLMs can significantly improve the efficiency, accuracy, and scalability of horizon scanning by extracting meaningful insights from large, unstructured datasets. The results highlight the potential of LLM-driven foresight to enhance regulatory preparedness, guide compliance strategies, and inform policy design in an increasingly complex and dynamic regulatory environment. 

Sustainable Financial Products and UK Pension Schemes

Sustainable financial products have gained significant traction in the financial world as climate change and social responsibility concerns continue to dominate public discourse. In the UK, Environmental, Social, and Governance (ESG) and sustainability considerations have been steadily gaining attention as both financial product designs and risk management tools.

Economic trends, regulations, and soft laws have been reactive over the last decade to growing transparency and demands for accountability (Palea, 20221; Escrig-Olmedo, Muñoz-Torres, Fernandez-Izquierdo, 20132).

This paper explores the growing role of sustainable financial products in the UK’s Defined Contribution (DC) pension schemes. It highlights key challenges and opportunities, focusing on the interplay between sustainable investment products, pension dashboards, Fintech, and institutional perspectives.

Mapping ESRS Disclosure Datapoints to Relevant Datasets

The integration of geospatial data into sustainability reporting frameworks addresses challenges related to inconsistent and outdated Environmental, Social, and Governance (ESG) information. This third white paper from the Financial Regulation Innovation Laboratory (FRIL) explores the application of geospatial data in enhancing the European Sustainability Reporting Standards (ESRS). By aligning geospatial datasets with specific ESRS disclosure requirements, the study provides a foundation for corporations conducting double materiality assessments, auditors validating disclosures, and third parties—such as financial institutions and environmental organisations—performing due diligence.

Geospatial data can be applied at the asset level (e.g., factories) or aggregated using a bottom-up approach linked to financial ownership, improving transparency and comparability across companies, sectors, and regions. However, the study finds that only 7% of ESRS datapoints can be externally validated due to the dependence on proprietary company information. Despite this limitation, different stakeholders benefit from distinct datapoints: investors may prioritise datapoints linked to external risks such as flooding or greenhouse gas emissions, while water-focused non-governmental organisations may emphasise hydrological indicators.

The EU Omnibus package (February 2025) introduces significant changes to ESRS and corporate sustainability reporting. These include a reduction in in-scope companies (80% fewer under the Corporate Sustainability Reporting Directive), limited value chain coverage, and fewer required datapoints, which may lead to a data gap and reduced transparency. However, the shift towards quantitative over qualitative datapoints presents a critical opportunity for geospatial data to bridge this gap, offering independent, real-time, and scalable insights for ESG reporting.

Furthermore, the revision of assurance requirements under the Omnibus package raises concerns about data verification and reporting accuracy. Given these regulatory shifts, integrating satellite- derived data into sustainability reporting frameworks could enhance objectivity, comparability, and reliability. Future regulations should embed geospatial data as a core element to strengthen the integrity and effectiveness of sustainability disclosures in the EU and beyond.

Generative AI for Simplified ESG Reporting in Financial Services 

We demonstrate the potential for Generative AI to simplify Environmental, Social, and Governance (ESG) reporting in financial services. Banking and financial institutions are required to comply with ever more stringent and demanding ESG related compliance requirements. A lack of mandatory, universally enforceable sustainable finance standards and guidelines makes effective ESG reporting across industries and countries difficult for financial institutions.

Vast amounts of data processing are required, spanning structured quantitative numerical data and unstructured qualitative textual data. Generative AI has the potential to deliver an innovative solution to this ESG reporting challenge through identifiable capabilities in decision support, including document summarisation; data visualisation; individual and multiple company analytics; and customised report generation. Furthermore, several technical features allow organisations to customise Generative AI systems to meet bespoke business requirements and information technology constraints.

These technical features include response speed and agility; multiple version choice and algorithmic support; user friendly interfaces; scalability and upgradability. In the use case demonstration, we show how a Large Language Model (LLM) can be used to generate responses to a set of common analyst questions pertaining to ESG using single and multiple annual report sources.

This use case brings to life the potential for Generative AI in simplifying compliance in respect of ESG reporting. We then bring together LLM and cutting-edge large Vision Model (LVM) capability to move from text-based prompting to verbal-based prompting for the ESG reporting exercise. We show that this integrated language-vision approach leads to enhancements in performance compared to a sole LLM approach. Indeed, we demonstrate that placing emphasis on key words within the verbal prompts generates more targeted responses from the LLM.