《RAND：2024生成式人工智能对信息完整性的威胁和潜在的政策应对措施（英文版）（24页）.pdf》由会员分享，可在线阅读，更多相关《RAND：2024生成式人工智能对信息完整性的威胁和潜在的政策应对措施（英文版）（24页）.pdf（24页珍藏版）》请在三个皮匠报告上搜索。

1、The advent of generative artificial intelligence(AI)and the growth of broader AI capabilities have resulted in many questions about the technologys impact on the body politic and the role of information and trust in a func-tioning democracy.Even before the 2022 generative AI breakthrough when OpenAI

2、 released ChatGPT and DALL-E,the United States and other nations suf-fered from what RAND researchers have referred to as truth decay,or the decline of the role of facts and analysis in public life.1 However,generative AI opens up even newer avenues of both malicious use and unwitting misuse of info

3、rmation,which can lead to sweeping implications for election cycles,the empowerment of danger-ous nonstate actors,the spread of both misinformation and disinformation,and the potential to undermine electoral processes.2 The purpose of this paper is to provide policymakers and scholars with a brief a

4、nd high-level review of potential threats that generative AI might pose to a trust-worthy information ecosystem.3 We offer a summary of policy initiatives that could mitigate these threats.In this paper,we avoid offering specific recommendations TODD C.HELMUS,BILVA CHANDRAGenerative Artificial Intel

5、ligence Threats to Information Integrity and Potential Policy Responses Expert InsightsPERSPECTIVE ON A TIMELY POLICY ISSUEApril 20242AbbreviationsAIartificial intelligenceEUEuropean UnionLLMlarge language modelNISTNational Institute of Standards and TechnologyNPDnonconsensual pornographic deepfakef

6、or policy initiatives,but we try to summarize the strengths and limitations of major proposals.We conducted this review by examining a variety of published resources on generative AI threats to the information space and papers that highlight potential policy options.4 For a more com-prehensive take