1、PARTNERSHIP ON AIAI Agents and Global Governance:Analyzing Foundational Legal,Policy,and Accountability Tools1AI Agents and Global GovernanceAnalyzing Foundational Legal,Policy,and Accountability Tools Talita DiasContentsExecutive Summary 3Introduction 51.Cross-border Harms 8International law 8Non-b

2、inding global norms 11Global accountability mechanisms 122.Human Rights Impacts 12International law 13Non-binding global norms 14Global accountability mechanisms 14Conclusion:Gaps and Recommendations 15Endnotes 20PARTNERSHIP ON AIAI Agents and Global Governance:Analyzing Foundational Legal,Policy,an

3、d Accountability Tools3Executive Summary Artificial intelligence(AI)agents are advanced computer programs that can set their own goals,make plans,and act in ways that are not entirely predictable.Unlike chatbots,they are anticipated to work with a high degree of autonomy to tackle increasingly compl

4、ex tasks.While not yet widespread,some AI agents can take action in external environments by inter-acting with different systems,such as by calling tools or writing and running computer code.These capabilities are expected to make AI agents useful for different tasks for example,streamlining adminis

5、trative tasks or optimizing critical systems.But the ability to take autonomous action in external environments and directly affect software or hardware in unexpected ways could amplify well-known AI risks and create new ones.In a hypercon-nected world,many of these risks could cross national border

6、s or concern the international community as a whole.Potentially significant global risks include:1.Cross-border harms,such as electoral interference and critical infrastructure disruption.2.Human rights impacts,such as privacy breaches and free speech limitations.This brief explores how these two se