<p>Music therapy has emerged as a popular way to improve mental and physical health in recent decades. Music therapy is recognized for its efficacy in addressing serial mental concerns via regulating emotion. However, traditional music therapy still faces some issues, such as copyright disputes and personalisation. With the rapid development of artificial intelligence (AI) technology, generated music based on AI-generated content provides the potential to solve these issues. Nevertheless, while real music has been shown to affect people’s moods, little research focuses on whether AI-generated music can influence human emotions as effectively as real music. In this work, we explore the emotional effects of AI-generated and real music via electroencephalogram (EEG). In the experiments, we collect EEG recordings while subjects respectively listen to AI-generated and real music. We then analyze the emotional effect in the two scenarios via handcrafted EEG features. Results show that most emotions exhibit over 65% similarity between AI-generated and real music in EEG features, and all subjects’ fear responses exceed 70%.</p>