Language models, particularly transformer-based architectures like ChatGPT, have gained significant attention due to their ability to comprehend and generate human-like text. These capabilities are leveraged to retrieve Green Image Damaging (GID) information about a randomized sample of about 400 of the largest production companies. For each sample company prompts to ChatGPT are used to discover and retrieve information of five company topics: CO2 compensation, greenwashing, environmental scandals, noncompliance with environmental legislation and standards, and legal actions related to environmental violations. Through corresponding data analysis, the study explores differences in extents of GID information for regions and industry sectors using the NACE classification scheme. Based on the extent of obtained information the sample is divided into companies where GID information is discovered and companies without GID information. The two groups are compared in terms of company size and ESG scores. Among other results the data analysis suggests that companies with GID information are larger and have significantly better ESG scores than the companies without GID information.