The aim of this study was to design and validate an AI-based predictive model to identify greenwashing patterns in Persian online marketing content. Greenwashing, as a challenging phenomenon in green marketing, involves exaggerated and misleading environmental claims by organizations that lead to reduced consumer trust and destruction of healthy competition.
This was an applied-developmental study and 180 Persian online advertisements from the food industry were collected and labeled by experts. After data preparation with natural language processing, lexical, rhetorical, and syntactic features were extracted and a random forest algorithm was trained.
The results showed that the model has a high ability to identify greenwashing with an accuracy of 88.9% and an AUC of 0.957. Ambiguous environmental keywords and nature-based metaphors had the highest predictive role. Qualitative analysis showed that companies use emotional and exaggerated vocabulary without real support.
This study showed that greenwashing in Persian advertisements follows specific linguistic patterns and can be detected by artificial intelligence. The developed model is a practical tool for consumers and regulatory agencies. It is suggested that future research should develop more comprehensive frameworks by expanding the database and adding new variables.