Law firms face growing pressure to manage the risks of generative AI as courts sanction attorneys for submitting filings built on hallucinated case citations. Insurance professionals and legal technology experts interviewed by The Daily Record offered five practical steps for firms adopting AI tools.
Write the policy down
Al Roberts, who heads professional liability underwriting for Berkley Select, said firms should establish formal AI-use policies rather than allowing lawyers and staff to determine their own practices. He encourages business leaders to set firmwide guidelines identifying approved AI tools, permissible uses and required review procedures.
A written policy gives lawyers clear guardrails before they experiment with AI on client matters. It also gives insurers a clearer picture of a firm's risk profile when questions arise.
Verify everything
Chris Newbold, chief operating officer and executive vice president of ALPS, a Montana-based lawyers' professional liability insurer, emphasized that lawyers remain responsible for verifying their work, even when AI is used. Lee Norcross, president of L Squared Insurance Agency, warned that attorneys can be sanctioned or face professional discipline for failing to verify AI-generated work.
The experts advise treating AI output as a first draft, not a final product. Independently verify every citation, quotation, factual statement and legal conclusion before relying on it in client work or court filings.
Protect client confidentiality
Roberts said firms should consider closed or controlled AI environments designed to prevent client information from being disclosed, retained for training or exposed outside the firm. Newbold cautioned lawyers against putting confidential information into free, publicly available AI platforms.
That distinction matters. A public chatbot might store or train on whatever a lawyer pastes into it, while a controlled enterprise deployment can keep client data inside the firm's own systems.
Train lawyers and staff
Tom Martin, creator of CiteCheck AI, said lawyers need to educate themselves about how AI systems work before using them. They need to understand AI's capabilities and limitations, applicable ethics requirements and firm policies. Roberts noted attorneys also need to understand any ethics requirements governing AI use in each jurisdiction where they practice.
Training should not stop at a single lunch-and-learn session. Firms that build ongoing education into their workflow - whether through internal programs or structured courses like the AI Learning Path for Paralegals - give staff a repeatable way to stay current as the tools change.
Professional judgment cannot be delegated
AI can improve efficiency, but it cannot replace a lawyer's independent legal judgment, the experts said. Attorneys - not the technology - remain responsible for the accuracy of their work product. Newbold said lawyers continue to "own the judgment, the verification and the work product" when they use AI.
That means the lawyer who signs the brief bears the consequences if an AI-generated citation turns out to be fabricated. No software license or vendor disclaimer transfers that accountability.
Why this matters for legal professionals
For practicing lawyers, paralegals and firm administrators, the practical takeaway is straightforward: adopt AI deliberately or risk sanctions, malpractice exposure and client trust. A written policy, verified outputs and proper training - such as the AI for Legal resources covering research and compliance automation - turn AI from a liability into a defensible tool. The firms that treat AI as a supervised assistant rather than a replacement for judgment will be the ones that use it safely.
Your membership also unlocks:
