AI Agents and Conflicts of Interest

A new paper by Ella Genasci Smith, Victor Y. Wu, and Jennifer King addresses the governance of agentic AI, focusing on potential conflicts of interest between AI developers and users. It highlights that the rise of non-human agency necessitates an overhaul of existing liability paradigms, as current legal systems are ill-equipped to handle the new risks. The authors propose policy recommendations such as applying the duty of loyalty to AI agents, implementing digital agent identifiers, enacting federal privacy legislation, and establishing agent adverse incident reporting.

• The spread of non-human agency will require an overhaul of liability paradigms.
• Current legal systems do not adequately address the risks posed by AI agents.
• The concept of an ‘agentic economy’ requires rethinking AI accountability.
• The paper discusses applying the duty of loyalty to AI agents.
• Other policy recommendations include digital agent identifiers, federal privacy legislation, and adverse incident reporting.
  https://bender.layer3.press/articles/832ed8a0-50f6-4d81-98f8-4e8c67506201