To achieve global sustainability goals, enterprises are obliged to declare sustainability of their economic activities as part of EU Taxonomy reporting. Due to a lack of capacity and expertise, SMEs are unable to adequately fulfil this reporting requirement. A significant degree of automation is required to enable SMEs to efficiently comply with reporting. Recent advances in generative AI entails potentials to overcome several technical challenges inter alia i) heterogeneity of data, ii) necessary semantic intelligence, iii) usability requirements, and iv) assuring quality and reproducibility of automatically generated reports. This paper proposes a novel AI framework for automating sustainability reporting, thus assisting sustainability managers. It uses a hybrid approach incorporating knowledge graphs (KG) and large language models (LLM). Firstly, the EU taxonomy is converted into a Taxonomy-KG that specifies required KPIs and identifies target sources for data collection, thus retrieving necessary information for feeding KPIs. Subsequently, the Taxonomy KG guides an LLM to extract relevant information from corporate data, thereby enabling the EU Taxonomy reporting. By feeding the KPIs, a sustainability manager chatbot automatically generates an EU taxonomy report.