The rapid advancement of generative AI has created a critical gap between technological innovation and responsibility frameworks. This article examines the comprehensive challenges posed by AI systems that can autonomously generate content and make decisions affecting crucial social domains. We analyze the failure of traditional accountability mechanisms in addressing AI's emergent behaviors and ``black box'' characteristics, and propose a multi-dimensional approach to responsibility allocation. The analysis covers five key areas: the primary responsibilities of technology developers throughout the AI lifecycle, the necessary paradigm shifts in legal frameworks including new concepts of algorithmic accountability, the transformation of users from passive recipients to active participants in the responsibility chain, the imperative for global collaborative governance transcending national boundaries, and the dynamic nature of responsibility boundaries that must evolve with technological advancement. We argue that establishing robust responsibility frameworks is not about constraining innovation but providing sustainable institutional support for AI development. The boundaries of AI accountability must be continuously negotiated through the dynamic interplay of collective human wisdom, balancing efficiency with fairness, innovation with security, and development with sustainability, to ensure that AI technology ultimately serves as a positive force advancing human civilization.