Formalizes AI agent governance as 'policies on paths,' moving from static prompts to runtime enforcement of complex legal and safety constraints.
March 18, 2026
Original Paper
Runtime Governance for AI Agents: Policies on Paths
arXiv · 2603.16586
The Takeaway
Current agent safety relies on fragile system prompts; this framework treats the agent's execution path as a mathematical object that can be audited and restricted in real-time. This is a critical step for deploying agents in high-stakes environments like finance or law.
From the abstract
AI agents -- systems that plan, reason, and act using large language models -- produce non-deterministic, path-dependent behavior that cannot be fully governed at design time, where with governed we mean striking the right balance between as high as possible successful task completion rate and the legal, data-breach, reputational and other costs associated with running agents. We argue that the execution path is the central object for effective runtime governance and formalize compliance policie