Visibility into AI Agents
This work addresses governance challenges for stakeholders in commercial, scientific, governmental, and personal domains, but it is incremental as it builds on existing risk frameworks without introducing new methods.
The paper tackles the problem of societal risks from AI agents by proposing and analyzing three categories of measures—agent identifiers, real-time monitoring, and activity logging—to increase visibility into their use, discussing implementations across deployment contexts and implications for privacy and power.
Increased delegation of commercial, scientific, governmental, and personal activities to AI agents -- systems capable of pursuing complex goals with limited supervision -- may exacerbate existing societal risks and introduce new risks. Understanding and mitigating these risks involves critically evaluating existing governance structures, revising and adapting these structures where needed, and ensuring accountability of key stakeholders. Information about where, why, how, and by whom certain AI agents are used, which we refer to as visibility, is critical to these objectives. In this paper, we assess three categories of measures to increase visibility into AI agents: agent identifiers, real-time monitoring, and activity logging. For each, we outline potential implementations that vary in intrusiveness and informativeness. We analyze how the measures apply across a spectrum of centralized through decentralized deployment contexts, accounting for various actors in the supply chain including hardware and software service providers. Finally, we discuss the implications of our measures for privacy and concentration of power. Further work into understanding the measures and mitigating their negative impacts can help to build a foundation for the governance of AI agents.