Five Rules to Reduce AI Liability Risk in Law Firms

Five Rules to Reduce AI Liability Risk in Law Firms

Law firms adopting generative AI face mounting liability concerns as attorneys receive professional sanctions for citing cases that AI systems fabricated. Insurance underwriters and legal technology specialists have outlined five concrete practices to help firms minimize malpractice exposure while using the technology.

Why it matters

The recommendations come as legal AI use accelerates without corresponding guardrails. Lawyers have already faced court sanctions for submitting AI-generated briefs containing nonexistent case citations, creating precedent that professional responsibility extends to verifying machine output. For firms without formal policies, individual attorneys are making ad hoc decisions about AI use that could expose the entire practice to liability.

Establish written firmwide policies

Al Roberts, who leads professional liability underwriting at Berkley Select, said firms should implement formal AI-use policies rather than leaving decisions to individual lawyers. These policies should specify which AI tools are approved, define permissible applications, and mandate review procedures before any AI-assisted work reaches clients or courts.

The absence of clear guidelines creates inconsistent practices across a firm and leaves attorneys uncertain about acceptable use boundaries.

Verify every AI output independently

Chris Newbold, chief operating officer at ALPS, a Montana-based legal malpractice insurer operating in Maryland, emphasized that attorneys retain full responsibility for work accuracy regardless of AI involvement. Lee Norcross, president of Michigan-based L Squared Insurance Agency, warned that lawyers face sanctions and professional discipline for failing to verify AI-generated content.

Both experts recommend treating AI output as preliminary drafts requiring human review. Attorneys must independently confirm every citation, quotation, factual assertion, and legal conclusion before incorporating AI-generated material into client deliverables or court filings.

Use closed AI environments for confidential data

Roberts advised firms to deploy closed or controlled AI systems designed to prevent client information from being disclosed externally, retained for model training, or exposed beyond the firm's boundaries. Newbold similarly cautioned against entering confidential information into free, publicly available AI platforms that may not protect attorney-client privilege or work product doctrine.

Train staff on AI capabilities and limits

Tom Martin, creator of CiteCheck AI, said lawyers must educate themselves about how AI systems function before deploying them. Training should cover the technology's capabilities, its limitations—including the tendency to generate plausible but false information—and applicable ethics requirements.

Roberts noted that attorneys need to understand AI-related ethics rules in every jurisdiction where they practice, as requirements vary by state.

Maintain professional judgment authority

While AI can enhance efficiency, it cannot substitute for a lawyer's independent legal judgment, the experts agreed. Attorneys remain accountable for their work product's accuracy. As Newbold put it, lawyers continue to "own the judgment, the verification and the work product" when using AI tools.

These recommendations were detailed in reporting by The Daily Record.