1. Home
  2. Centers and Institutes
  3. Information Law Institute
  4. Research
  5. AI and Agency Law

AI and Agency Law

AI is now a part of everyday life, and “AI agents” operate autonomously in complex domains. Predating these artificial agents by hundreds of years, agency law governs the relationship between a principal and their trusted agent in many domains, including health, finance, legal services, and business management. According to agency law, agents owe fiduciary duties of loyalty and care to their principals. We research how AI systems can be compliant with these duties in contexts where they already exist. We also research how expanded fiduciary duties can address longstanding problems with the digital economy, such as conflicts of interests that arise with platforms, and incomplete contracting. This work is both actionable today with current implementations of AI, and a way to manage AI alignment in the long term to forestall catastrophic risks.

Institutionalizing Fiduciary AI Standards

We are documenting the variety of institutional arrangements that are used to standardize and enforce fiduciary responsibilities currently, and analyzing which of these arrangements is best suited to the realities of AI. Special attention will be paid to AI pipelines and the limits of regulating opaque and complex systems. What interventions can be made, at what stage in the development and deployment process, and by whom, for standardization to be effective?

Fiduciary Data Intermediary AI Design

Anticipating new architectures for fiduciary web services, as well as EU rules establishing fiduciary duties for data intermediaries, we apply computational economics techniques to guide how fiduciary duties apply to AI systems with multiple principals and interactions with third parties.

Fiduciary AI and the Incompleteness of Consent

Consider one root of the rationale for fiduciary AI: that human users cannot meaningfully consent to a contract with an AI performing tasks beyond their comprehension. By mathematically clarifying this root problem, we will both solidify the case for fiduciary AI and enumerate the powers needed by a regulator to adequately enforce agency duties.

Selected Research
  • Peter Hall, Olivia Mundahl, and Sunoo Park, The Pitfalls of “Security by Obscurity” and What They Mean for "Transparent AI.," 39 Proc. AAAI Conference on A.I., 28042, (2025) (link).
  • Noam Kolt, Michal Shur-Ofry, and Reuven Cohen, Lessons from complex systems science for AI governance, 6 Patterns Aug. 8 2025 (link).
  • Ignacio Cofone and Warut Khern-am-nuai, The Overstated Cost of AI Fairness in Criminal Justice, 100 Ind. LJ 1431 (2024) (link).
  • Kate Crawford and Jason Schultz. The Work of Copyright Law in the Age of Generative AI, 94 Grey Room 56 (2024) (link).
  • Katja Langenbucher, AI Judgment Rule(s), 2024 U. Chi. L. Rev. Online 1 (link).
  • Katja Langenbucher, Ownership and Trust-A Corporate Law Framework for Board Decision-making in the Age of AI, (European Corp. Gov. Inst., Law Working Paper 758, 2024) (link).
  • Michal Shur-Ofry, Multiplicity as an AI Governance Principle, 100 Ind. L.J. 1527 (2024) (link).
  • Michal Shur-Ofry, A Networks-of-Networks Perspective on AI policy, 2024 Network L. Rev. 212 (link).
  • Jill Walker Rettberg, et al., An AI Society, 40 Iss. Sci. & Tech. Winter 2024, at 77 (link).
  • Jake Karr and Jason Schultz, The Legal Imitation Game: Generative AI's Incompatibility with Clinical Legal Education, 92 Fordham L. Rev. 1867 (2024) (link).
  • Sebastian Benthall and David Shekman. Designing Fiduciary Artificial Intelligence, EAAMO '23: Proc. of the 3rd ACM conference on Equity and Access in Algorithms, Mechanisms, and Optimization 1 (2023) (link).
  • Evan Selinger, Brenda Leong, and Albert Fox Cahn, AI Audits: Who, When, How... or Even If? in Collaborative Intelligence, MIT Press (Mira Lane, Arathi Sethumadhavan eds., 2023) (link). 
  • Aarohi Srivastava et al., Beyond the Imitation Game: Quantifying and Extrapolating the Capabilities of Language Models, Transactions on Machine Learning Research, May 1 2023 (link). 
  • Sebastian Benthall and Jake Goldenfein, Artificial Intelligence and the Purpose of Social Systems, AEIS '21: Proc. of the 2021 AAAI/ACM Conference on AI, Ethics, and Society, 3-12 (2021) (link).