Abstract: Regulating and governing generative artificial intelligence is fundamentally crucial for forstering its broad adoption and continuous innovation. Eusuring the technology is responsible, which is answerable for human interests, has emerged as a essential requirement for its advancement. The primary goal of establishing a responsibility mechanism for generative artificial intelligence is to bring a positive impact to human, based on a "human-centered" logic. At the societal level, the responsibility mechanism faces challenges from conflicted values; at the technical level, it confronts challenges of algorithmic opacity; and within the generative artificial intelligence industry, it is challenged by unregulated research and application. It is imperative for humanity to construct thr pathways to the responsibility mechanism of generative artificial intelligence, including promoting human-machine integration to guide technology toward benevolence, deconstructing algorithmic opacity to enhance technical transparency, optimizing government governance standards to strengthen resilient regulation, balancing interests, and reinforcing the social responsibility of artificial intelligence enterprises.