生成式人工智能网络谣言法律责任研究

Research on Legal Liability of Network Rumors Generated by Artificial Intelligence Networks

摘要: 随着以ChatGPT、Sora为代表的生成式人工智能技术的爆发式发展，网络谣言的生成与传播机制发生了质变。生成式人工智能技术不仅极大地降低了网络谣言的制造成本，更通过深度伪造等方式使网络谣言内容呈现前所未有的高度逼真性，借助算法推荐和社交机器人实现精准化、智能化的裂变式传播，构成了严峻挑战。现行法律规制在应对此类谣言时面临诸多困境，尤其是在责任主体认定和归责原则适用上存在模糊地带。当谣言的直接生产者是算法而非人时，如何在技术开发者、服务提供者和使用者之间分配责任成为难题。基于“控制力”理论，本文认为，应重构一种以控制力为核心的多元、分层责任体系。在归责原则上，应以过错责任为基础，并针对不同主体和场景引入过错推定和严格责任作为补充。通过明确各方主体的注意义务，并类推适用“通知–必要措施”规则，可以在鼓励技术创新的同时，有效遏制网络谣言的危害，实现技术发展与社会安全的协调发展。

Abstract: With the explosive development of generative AI technologies represented by ChatGPT and Sora, the mechanisms of rumor generation and dissemination have undergone fundamental transformations. These technologies not only drastically reduce the cost of creating online misinformation but also enable unprecedentedly realistic content through deepfake techniques. Leveraging algorithmic recommendations and social bots, they facilitate targeted, intelligent viral propagation, posing significant challenges to legal frameworks. Current regulatory systems face multiple difficulties in combating such rumors, particularly regarding ambiguous liability determinations and attribution principles. When algorithms rather than humans directly generate misinformation, allocating responsibility among tech developers, service providers, and users becomes particularly complex. Grounded in the “control theory”, this study proposes reconstructing a multi-tiered liability framework centered on control capacity. The attribution principle should adopt fault-based liability, supplemented by presumed fault and strict liability mechanisms tailored to different entities and scenarios. By clarifying each party’s duty of care and applying the “notice-necessity” rule through analogy, we can effectively curb the harm caused by online rumors while encouraging technological innovation, achieving harmonious development between technological progress and social security.