What steps can lawyers take to reduce AI-related liability risk?

Why Risk Reduction Has Become a Priority

As AI tools have become a routine part of legal research, drafting, and document review, so has awareness of the liability risks they can introduce when their output isn’t properly verified. Rather than avoiding AI tools altogether, many firms have instead focused on building practical safeguards that let them capture the efficiency benefits of AI while reducing the chance that an unverified error leads to a sanctions order or a malpractice claim.

Independent Verification as the Core Safeguard

The single most consistently emphasized safeguard across bar guidance and law firm risk management materials is independent verification: checking any AI-generated case citation, statutory reference, or factual claim against a reliable, authoritative source before relying on it in client work or a court filing. This step directly addresses the most well-documented AI risk in legal practice — fabricated but plausible-sounding content, including nonexistent case citations, generated by AI tools with no reliable indication to the user that anything is wrong. Treating AI output as a draft requiring confirmation, rather than a finished, trustworthy answer, is the mindset shift most guidance recommends.

Written Firm Policies and Approved Tools

Beyond individual diligence, many firms have found it valuable to adopt written policies governing AI use — specifying which tools are approved for which tasks, what data can and cannot be entered into a given tool (particularly regarding client confidentiality), and what review steps are mandatory before AI-assisted work product is finalized or filed. Clear, written policies help ensure consistent practice across a firm rather than leaving each attorney to develop their own ad hoc approach, and they also give firms something concrete to point to if their AI practices are ever questioned.

Documentation and Training

Maintaining some record of verification steps taken — even something as simple as noting that citations were checked against a specific legal database — can help demonstrate reasonable diligence if a piece of work is later challenged. Regular training, especially for newer associates and staff who may be more inclined to trust AI output at face value, also helps reinforce awareness of the specific failure modes AI tools can exhibit, including the risk of confidently presented but fabricated content.

Staying Current With External Guidance

Because this is a fast-moving area, firms benefit from monitoring evolving guidance from their state bar association, relevant court rules on AI disclosure in filings, and general professional responsibility developments. What counts as adequate diligence is likely to keep evolving as AI tools and their known risks become better understood across the profession.

Bottom Line

Lawyers can meaningfully reduce AI-related liability risk by independently verifying AI-generated content before relying on it, adopting clear firm-wide policies on approved tools and required review steps, documenting diligence, and staying current with evolving bar and court guidance.

Go deeper

• How does the government evaluate and procure AI systems differently from other software?