生成式人工智能产品的内容安全风险及监管路径

Content Security Risks and Regulatory Paths of Generative Artificial Intelligence Products

ChatGPT的问世引发了社会各界的广泛关注和讨论,生成式人工智能产品会对内容安全产生哪些影响、如何看待这些影响是其中的关键问题。解答这些问题需要破除充满科幻色彩的想象,理性剖析生成式人工智能技术的本质特征及可控性。研究认为,生成式人工智能产品在内容安全上具有可控性,但同时亟待通过完善人工智能基础性法律法规、实施分类管理、开展安全评估与安全审计、强化社会教育等治理方式,找到生成式人工智能产品安全与发展的最优平衡。

The emergence of ChatGPT has attracted widespread attention and discussion in society.To understand the key issues surrounding the impact of generative artificial intelligence products on content security,it is necessary to dispel the sci-fi imaginings and rationally analyze the essential characteristics and controllability of these technologies.This paper argues that generative artificial intelligence products have controllability in terms of content security,but at the same time,it is urgent to find the optimal balance between security and development through comprehensive legal regulations on artificial intelligence,implementation of classification management,security assessments and auditing,and strengthening user education.