Abstract

When an AI tool deployed in legal practice produces an erroneous output — a hallucinated citation, a mischaracterized precedent, a flawed contract clause — the question of liability is neither simple nor settled. The attorney who relied on the tool, the firm that deployed it, and the vendor who built it may each bear some portion of responsibility. Existing tort doctrine, professional responsibility rules, and contractual frameworks provide incomplete and sometimes contradictory guidance on how that responsibility should be allocated.

This paper examines the liability landscape when AI tools fail in legal practice. We analyze attorney malpractice exposure under the competence and supervision obligations of the Model Rules, the enforceability of vendor liability disclaimers under UCC Article 2 and common law unconscionability doctrine, and the potential for direct claims against AI vendors under emerging product liability theories. We argue that the current framework systematically under-deters AI vendor negligence while over-exposing attorneys who exercise reasonable professional judgment in deploying AI tools.

We propose a tiered liability framework that allocates responsibility based on the nature of the AI failure, the reasonableness of attorney reliance, and the adequacy of vendor disclosure. The framework draws on analogies from medical device liability, software product liability, and professional services malpractice to construct a coherent approach to AI tool failure in legal contexts.

Full Paper

This paper is published on the Social Science Research Network (SSRN). To read the full text, download the PDF, or cite this work, please visit the SSRN abstract page:

→ Read the Full Paper on SSRN (Abstract ID: 6741139)

Published: May 2026 — Authors: Austin, Morris & Das — View all research papers