Greenwashing can be defined as the mismatch between companies’ positive green communications and their activities that are counterproductive to the fight against climate change. Greenwashing companies can undermine the success of climate policies by hiding their climate footprints and prevent policymakers from realizing the need for further regulations. Thus, the identification of greenwashing should be done timely: this research sets out to find a semi-automated approach that can find evidence of greenwashing. The study of greenwashing is limited to the mismatch between companies’ communications on official websites and their lobbying activities in the European Union. Some steps of the design research approach were used namely the design development, demonstration, and evaluation (via a small-case evaluation) phases. The created approach measures companies’ communication levels of commitment to climate change using a GPT-based sentence classifier and then compares it to their levels of lobbying on climate change using keyword-based filtering on the EU Transparency Register, the official lobbying source for the EU. The proposed method is semi-automated at this stage and can inform a greenwashing appraisal, but it has issues with the accuracy of its analysis. This research partially fills the gaps in the greenwashing literature and advances the research in the use of GPT for classification purposes and in the field of climate lobbying in the EU. This research can teach the implementation of new EU regulations that aim at banning greenwashing: the more detailed the definitions of greenwashing that are used, the easier to find greenwashing automatically, and, therefore, the faster the enforcement of the ban can be.