Process model generation is still mostly a manual task. A vast amount of process-relevant information is captured in textual sources such as regulatory documents and process descriptions. Hence, (semi-) automatic extraction of process model information from textual sources and translation into process models based on graphical notations such as BPMN are ongoing and, with the advent of generative AI, increasingly performant. However, existing approaches for process model generation from text are often limited to control flow aspects and require precise process descriptions as input. Regulatory documents describing processes pose different challenges than process descriptions and are widespread, highly relevant and of increasing volume in organizations. In this work, we exploit a generative AI architecture to present an approach that can automatically generate BPMN 2.0 process models from regulatory texts such as the GDPR. The approach is evaluated on five use cases with regulatory text and corresponding process models from different domains and with different challenges. This way, we can demonstrate that the proposed approach can automatically generate reference models from regulatory requirement documentations. When compared to existing approaches, the proposed approach results in an improvement of both, syntactic and semantic process model quality.