Generative AI (GenAI) is transforming our world. However, GenAI poses significant challenges, including authorship attribution (AA) e.g. determining whether text was written by a human or GenAI. This research tackles the problem of AA in a novel way – creating ‘emotionally-aware’ AI to detect GenAI text. Our approach shows improvements in GenAI detection across domains, datasets, model sizes (generator and detector), and model architectures. In addition, we analysed the affective profile of human and GenAI text – revealing surprising similarities and differences. Finally, we developed a novel flexible framework for evaluating AI models – whether the models predict discrete or continuous values.