Generative AI Empowers Brain-Computer Interfaces: A Review-Perspective on Technical Realities and Future Visions

Generative artificial intelligence (AI) has recently emerged as a transformative paradigm for brain-computer interfaces (BCIs), with impact spanning hardware, data, algorithms and applications. This paper systematically analyzes the convergence trends and development prospects of generative AI and brain-computer interface (BCI) systems. In particular, by defining the notions of generative brain decoding and discriminative brain decoding, it examines how generative AI is reshaping the decoding paradigms of BCIs. Generative AI has shifted BCI decoding from conventional discriminative paradigms toward generative decoding, enabling the reconstruction of semantically rich outputs such as text, speech, and images. Through data augmentation and the generation of virtual samples, generative AI improves the quality of BCI data and mitigates fairness issues arising from data scarcity and population imbalance. In hardware design, generative methods accelerate the discovery of electrode materials and sensor architectures that enhance stability and biocompatibility. These advances expand BCI applications in medical rehabilitation, industrial systems, and consumer electronics. At the same time, the integration of generative AI introduces critical challenges in security, privacy, and ethics, particularly concerning the misuse of synthetic data and unauthorized cognitive inference. This survey reviews current progress, identifies persistent challenges, and outlines future directions for developing reliable and equitable generative AI-empowered BCIs.