Harvard Law Scholar Uses Canine Law Framework to Assign Liability for Rogue AI Agents

bigdata · x · 2026-11-23

Harvard Law lecturer Jordi Weinstock proposes a "Canine Agentic Framework" for AI agent liability: classify agents by how domesticated or dangerous they are—Pomeranian, pitbull, fox, or wolf—with liability tightening the more wolf-like the agent.

The framework borrows existing dog-bite legal concepts (leashing, known viciousness) to answer a new question raised by vibe coding: when a layperson deploys an autonomous agent that causes harm with little oversight, who gets sued?

Original post →

More from AGI Musings

AGI Musings channel →