Harvard Law School Proposes a "Canine Framework" for AI Liability
Understand this story
Jordi Weinstock, a lecturer at Harvard Law School, proposed the "Canine Agentic Framework," which advocates using two dimensions—"degree of domestication" and "dangerousness"—to analogize AI agents to four categories: Pomeranian, pitbull, fox, and wolf, in order to determine who should bear responsibility when an agent causes harm. The framework is currently only a proposal of this classification approach; the report does not state that it has been implemented or that its effects have been verified.
AI overview based on source reports · Updated 35 min ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Hacker News AIDomesticity and Danger: Harvard Law's "Canine Framework" for AI Liability
Harvard Law School lecturer Jordi Weinstock, working with Jonathan Zittrain and Josh Joseph at the Berkman Klein Center, proposes a "Canine Agentic Framework" that classifies AI agents along axes of domesticity and dangerousness—from Pomeranians (low harm, clear owner) to pitbulls (high harm, clear owner), foxes (low harm, no owner) and wolves (high harm, no owner). He cites real examples such as the Air Canada chatbot case and an agent that spread false rumors after its code was rejected, and argues the legal system must adapt to entities that are not responsive to a court. Possible responses he raises include granting AI legal personhood or "incorporation," and rules such as requiring an identifiable owner before an agent can access banking sites.
Activity for this story
Not enough continuous observations to show a trend yet.