DeepGreen introduces a dual-layer LLM system that analyzes corporate financial statements to detect misleading environmental claims.

• Uses preliminary keyword identification followed by semantic implementation analysis
• Quantifies greenwashing through the novel GreenImplement ratio metric
• Tested on 68 companies from China's A-share market with 204 financial statements
• Serves as a proactive monitoring tool for regulators and investors

This research enhances financial security by providing a systematic method to identify potential environmental misrepresentations before they harm stakeholders or lead to regulatory penalties.

DeepGreen: Effective LLM-Driven Green-washing Monitoring System Designed for Empirical Testing -- Evidence from China