ChatGPT模型的法律风险及应对之策  

Legal Risks of ChatGPT Model and Resolutions

作　　者：邓建鹏 朱怿成 DENG Jian-peng;ZHU Yi-cheng(Law School;Central University of Finance and Economics,Beijing 100081;Law School,Peking University,Beijing 100871)

机构地区：[1]中央财经大学法学院 [2]金融科技法治研究中心,北京100081 [3]北京大学法学院,北京100871

出　　处：《新疆师范大学学报（哲学社会科学版）》2023年第5期91-101,F0002,共12页Journal of Xinjiang Normal University(Philosophy and Social Sciences)

基　　金：中央财经大学新兴交叉学科建设项目“金融系统安全与区块链监管科技”(2021-03)的阶段性成果。

摘　　要：人工智能聊天机器人ChatGPT一经推出就引发广泛关注,其为社会带来巨大便利的同时,也存在诸多法律风险,主要包括知识产权归属争议,数据来源合法性与用以训练的数据质量问题,生成不良或违法信息,挑战学校及学术伦理规范,等等。对此,监管机构与立法机构应提供相应的规制对策,包括明确类ChatGPT人工智能生成内容的知识产权取得规则,构建合理的市场准入制度和严格的责任承担制度,加强全链条监管;研发机构应履行合规义务,推出学术抄袭检测模型,提高类ChatGPT人工智能系统服务人类的水准和安全性。ChatGPT,an artificially intelligent chatbot,has attracted worldwide attention since its launch.While it brings great convenience to society,there are also many legal risks,including disputes over the ownership of intellectual property rights,the legality of data sources and the quality of data used for training,the generation of undesirable or illegal information,and the problems of examination cheating and academic plagiarism.To avoid or resolve the above problems,regulators and legislators should provide corresponding regulatory countermeasures,which include clarifying the rules for obtaining intellectual property rights of ChatGPT-like AI-generated contents,building a reasonable market access system and a strict liability system,and strengthening the whole chain of supervision.R&D institutions should fulfill their compliance obligations and improve the level and safety of ChatGPT-like AI services for human beings.

关 键 词：ChatGPT 内容生成式人工智能 人工智能法律风险 规制对策 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]