生成式人工智能的刑事法律风险及其合规治理

The Criminal Legal Risks and Compliance Governance of Generative Artificial Intelligence

大数据时代,生成式人工智能迅猛发展,呈现出强大的创造性与适应性。但由于其运作机理及内生缺陷等原因,生成式人工智能在输入、运算及输出各个阶段,可能引发诸如危害国家安全,侵犯公民个人信息、知识产权、商业秘密等刑事风险。应当秉承前瞻性治理理念,坚持刑法的谦抑性,避免刑法过度介入,构建事前合规、事中监管、事后整改的三重分级合规治理体系,以期有效应对刑事责任风险,同时为人工智能技术的广泛应用提供法治保障。

In the era of big data,generative artificial intelligence has developed rapidly,showing strong creativity and adaptability.However,due to its operating principle and endogenous defects,generative AI may cause criminal risks such as endangering national security,infringing personal information,intellectual property rights,commercial secrets,and violating citizens’personal and property rights at various stages of input,calculation and output.The ides of forward-looking governance should be adhered to,the modesty of criminal law should be upheld,excessive involvement of criminal law should be avoided,and a triple-graded compliance governance system should be constructed with regard to compliance beforehand,supervision during the process,and rectification after the process,with a view to effectively addressing the risk of criminal liability,while providing rule of law safeguards for the wide application of AI technology.