以DeepSeek为代表的生成式人工智能信息证据能力问题研究

中图分类号］D915.3;TP18 [文献标识码］A [文章编号]1008-1763（2025）04-0131-09

Abstract：Generative Artificial Inteligence （AI） technologies，represented by DeepSeek，enable content generation in various forms through deep learning and big data analysis. This raises the possibility of utilizing such content in criminal proceedings，with the core issue being the evidentiary capacity （i.e.，admissibility）of such information.While information generated by Generative AI may serve as litigation evidence under certain circumstances，it must meet requirements such as reliability，relevance，and legality. As DeepSeek has been applied for a short period，its technical maturity remains insufficient，and its reliability requires further validation. Therefore，caution should be exercised in treating its outputs as litigation evidence. Generative AI offers advantages in improving evidence-processing efficiency，expanding evidence types，and supporting objective fact-finding. However，it also poses the risks related to authenticity verification，algorithmic transparency， privacy protection，and technical immaturity，which present challenges in determining its evidentiary capacity. Drawing on international experiences such as the EU Artificial Intelligence Act and the U.S. AI Risk Management Framework，this study proposes building a technology credibility mechanism，refining evidence rule systems，and enhancing judicial capabilities to address challenges posed by advancing AI technologies.

Key words： Generative Artificial Intelligence; DeepSeek;criminal proceedings;evidentiary capacity; technical credibility

一 引言

无论是《刑事诉讼法》还是《民事诉讼法》，均将“电子数据"作为证据种类之一，也就是在立法上承认电子数据的证据能力。（剩余16940字）