内容安全是指对信息内容的保护，以及信息内容符合政治、法律、道德层次的要求。人工智能的发展对内容安全产生了非常重要的影响。本文从基于人工智能的内容安全重大战略需求出发，梳理了国内外的研究现状与发展趋势，凝练了基于人工智能的内容安全的关键技术问题。研究提出，按照三步走的策略构建世界领先的基于人工智能的内容安全体系；在对抗性机器学习、可解释人工智能、混合增强智能、知识驱动的内容安全等方面开展技术创新突破，同时应注重政策法规和监管机制建设；建设面向内容攻防的网络靶场、面向舆情攻防的大规模社会系统模拟装置等内容安全重大基础设施。

Content security refers to the protection of information content and that the information content meets the requirements at political, legal, and moral levels. The recent development of artificial intelligence (AI) has had a very important impact on content security. In this article, we summarize the research status and development trends of AI-based content security in China and abroad based on the major strategic demand therefor, and presents the key technical issues regarding AI-based content security. This study proposes to build the world’s leading AI-based content security system through a three-step strategy. Innovation and breakthroughs should be made in areas such as adversarial machine learning, explainable AI, hybrid enhanced intelligence, and knowledge-driven content security. Meanwhile, the construction of policies, regulations, and regulatory mechanisms should be emphasized. Furthermore, major content security infrastructure such as cyber ranges for content attack and defense and large-scale social system simulation devices for public opinion attack and defense should be established.