Research
Advancing AI alignment, multi-agent cooperation, and governance frameworks
for frontier AI systems—bridging computer science, law, economics, and social theory.
Research Overview
My research addresses fundamental questions about how to build AI systems that can operate safely,
ethically, and beneficially within human normative frameworks. This work spans three interconnected areas:
normative AI alignment, multi-agent cooperation, and governance mechanisms for advanced AI.
I draw on legal theory, economics, game theory, and computer science to develop both theoretical frameworks
and practical solutions. Central to this work is the insight that effective AI governance and alignment
require understanding how humans construct and maintain normative order—from informal social norms to
formal legal systems.
How do we build AI systems that can navigate the complex, context-dependent normative systems that
govern human behavior? I explore how insights from economics and evolutionary theory can inform AI
alignment—focusing on how AI can learn to recognize, reason about, and operate within human rules, norms,
and values. This includes developing training environments and architectures that enable AI to become
"normatively competent."
Key Questions
- How can AI systems learn to understand and operate in human normative frameworks?
- What role can legal reasoning and contract theory play in AI alignment?
- How do we train AI to recognize normative infrastructure in its environment?
- What makes rules "legible" to learning agents, and why do silly rules help?
TMLR
2024
With Atrisha Sarkar, Andrei Muresanu, and others
Proposes an architecture enabling AI agents to learn, represent, and reason about social
norms in ways that support cooperation in multi-agent environments.
Science Advances
2023
With Aparna Balagopalan, David Madras, David H. Yang, Dylan Hadfield-Menell, Julia Stoyanovich
Shows that standard ML data labeling practices are inadequate when models are used to make
normative judgments about humans, demonstrating the need to distinguish between factual and
normative labels with implications for fairness.
PNAS
2022
With Raphael Köster, Dylan Hadfield-Menell, Joel Z. Leibo, and others
Demonstrates experimentally that arbitrary ("silly") rules help artificial agents learn to
comply with and enforce norms more effectively—a key insight for building normatively
competent AI.
As AI systems become more numerous and autonomous, understanding how they interact becomes critical. I study cooperation, coordination, and conflict among AI agents. This work draws on law, game theory, and mechanism design to address how we move beyond current approaches that lead to misaligned agent behavior—preventing harmful collusion while enabling productive cooperation in multi-agent systems.
Key Questions
- How can we build AI systems capable of grounded argumentation?
- How can we build AI systems that cooperate effectively with each other, and humans?
- What risks emerge from multi-agent interactions, and how can we mitigate them?
arXiv
2025
With Lewis Hammond, Alan Chan, and others
Analyzes risks that emerge specifically from interactions among multiple advanced AI
systems, including coordination failures, conflicts, and emergent behaviors.
TMLR
2025
With Alan Chan, Kevin Wei, and others
Proposes the technical and institutional infrastructure needed to support safe and
beneficial deployment of autonomous AI agents at scale.
NBER
2025
With Andrew Koh
Examines how economic principles apply to a world where AI agents transact, cooperate,
and compete, and what governance structures such an economy requires.
How do we govern AI systems that evolve faster than traditional regulatory frameworks can adapt? I develop novel governance mechanisms—including regulatory markets—that can keep pace with rapid AI advancement while managing risks. This work addresses the current shortcomings with brittle approaches to regulation that cannot keep pace with frontier AI model development.
Key Questions
- How can regulatory markets enable adaptive governance of rapidly evolving AI?
- What international institutions are needed to coordinate AI safety efforts globally?
- What technical infrastructure is necessary to enable safe, democratic AI development?
arXiv
2026
With Noam Kolt, Nicholas Caputo, Jonathan Zittrain, and others
Explores how legal rules, principles, and methods can address AI alignment challenges
through compliance mechanisms, interpretive frameworks, and structural blueprints for
reliable behavior.
Jurimetrics
2025
With Jack Clark
Proposes regulatory markets—where governments require AI companies to purchase regulatory
services from licensed private regulators—as a solution that overcomes the limitations
of both command-and-control regulation and industry self-regulation.
Science
2024
With Yoshua Bengio, Geoffrey Hinton, Andrew Yao, Dawn Song, and 20+ others
A landmark consensus paper describing extreme risks from advanced AI systems and outlining
a comprehensive plan combining technical research with adaptive governance mechanisms.
Science
2024
With Michael K. Cohen, Noam Kolt, Yoshua Bengio, Stuart Russell
Addresses the unique governance challenges posed by AI agents that can act autonomously
in the world, proposing regulatory approaches tailored to agentic systems.
arXiv
2023
With Markus Anderljung, Anton Korinek, and others
Proposes three building blocks for frontier AI regulation: standard-setting, registration
and reporting requirements, and compliance mechanisms tailored to AI's unique characteristics.
Explore Publications
View publications organized by research area, with full abstracts, citation information, and links to papers.
Browse Publications