主办单位:成都体育学院
ISSN 1001-9154 CN 51-1097/G8

Journal of Chengdu Sport University ›› 2025, Vol. 51 ›› Issue (1): 25-33.doi: 10.15942/j.jcsu.2024.01.03

Previous Articles    

Regulatory Paths for Applying Artificial Intelligence Technology to Large-scale Sports Events————Take Artificial Intelligence Video Surveillance as An Example

LI Haotian1, LI Yinhui2   

  1. 1. School of Law of Wuhan University, Wuhan Hubei 430064;
    2. Department of Physical Education of Zhongnan University of Economics and Law, Wuhan Hubei 430073
  • Received:2024-02-27 Published:2025-04-28

Abstract: The application of AI technology for security monitoring in large-scale sports events such as the Olympic Games, the World Cup and the Asian Games has triggered widespread concerns about data privacy, algorithmic discrimination and other risks. Given the limitations of the existing legislation in the digital domain to regulate such behaviors, it is necessary to reflect on the current path of regulating AI technology application based on risk prevention and based on layered governance. For intelligent video surveillance in sports event venues, the focus of regulation should be shifted from pre-defined risks to systematic monitoring of risks, and the center of regulation should be pushed forward from model-based hierarchical governance to the path of elemental hierarchical regulation. For the whole life cycle of intelligent video surveillance of large-scale sports events, a hierarchical regulation mechanism of “pre-training - fine-tuning and deployment” should be set up: in the data processing phase of pre-training, the dichotomy of labelled data and de-labelled data is maintained to dynamically improve the list of types of data related to the rights and interests of the state, enterprises and individuals, and the dynamic assessment of re-identification probability is carried out based on the degree of data significance and the degree of de-labelling. Based on the degree of data importance and de-identification, the probability of re-identification is dynamically valuated and information is disclosed. During the algorithm design and model development phases of pre-training, obligations related to managing risk, monitoring critical incidents, performing model evaluation and adversarial testing are constructed. In the fine-tuning and deployment phase, differentiate the functions of automatic classification and triage, automatic warning and prediction, and automatic forensics and evidence fixing for the intelligent monitoring system, and construct corresponding algorithmic auditing, algorithmic filing, disclosure, and interpretation mechanisms.

Key words: 2024 Paris Olympics, large-scale sports events, intelligent video surveillance, hierarchical regulation

CLC Number: