The AI mistake problem no lawyer can ignore

While all modern technology has experienced exponential growth, going from nascent ideas to ubiquitously used in everyday life, none more so than Artificial Intelligence (AI). ChatGPT was released in November 2022 and just three years later, almost everyone, lawyers included, is using some form of AI whether in their work or personal lives.

With lawyers, AI can be used to conduct document review, draft legal research memoranda, and draft initial documents, such as written discovery or even briefs. However, using AI for tasks traditionally performed by lawyers comes with great risk. So, what happens when something goes wrong?

The theory behind AI was to be able to replicate human intelligence in a machine. So, when trying to mitigate a claim, you must treat AI as a human. And humans are not infallible and are prone to mistakes. AI is as well.

While it may seem obvious, the first step to claim mitigation for AI-generated work-product is to always check its work. AI is not a shortcut to the end, it is a tool that helps you work faster. So, if you ask AI to conduct document review, have it create a summary that you can cross-check with the documents. For example, if you ask AI to perform a page-line summary of a deposition transcript, check to ensure the summary accurately reflects the deposition testimony. If you ask AI to draft written discovery, proofread and add case-specific requests. If you ask AI to perform legal research or draft a brief, check every single legal citation generated.

Further, AI is not going to identify quotations from an adverse party's deposition that would be ideal fodder for an opening statement. It's not going to create a case-specific request for admission that would be determinative. It's not going to add your voice to a brief, where a judge who is familiar with your work can hear you argue when they read it. The first step to claim mitigation is prevention and doing your due diligence is the best thing you can do to prevent mistakes when using AI.

But everyone makes mistakes--whether they are lawyers, AI or lawyers who failed to catch a mistake made by AI. Some mistakes are immediately apparent, others lie dormant for years--and while some are harmless, others carry serious consequences. When mistakes carry consequences, they must be repaired immediately, with an intelligent and effective solution to prevent them from escalating; while no one welcomes errors, the best course is to contain the damage to protect the client--and, in turn, yourself.

A claim (or potential claim) can arise from any mistake in any legal matter, dispute, case or representation, irrespective of its scope. For example: in trusts and estates (AI drafting the trust or estate plan, but AI did not account for the tax penalties), in family law (AI mistakes in financial tracing during a dissolution proceeding), during litigation (AI failing to calendar a deadline), during transactions (AI failing to include key provisions in documents), or in information data breach (AI failing to identify documents as attorney-client privileged).

Once a mistake is identified, the first step is to rectify it before consequences attach. Have the documents been signed yet? Has the trust been funded? Has the final deadline to submit trial exhibits passed? Can you file a Code of Civil Procedure § 473 motion? Can you claw back any attorney-client privileged documents? If yes, take all such steps to repair the mistake.

If, however, the mistake is not fixable, it must be reported to your insurer. While there are certainly concerns about how to make a claim and when to make it, generally, the earlier you notify your insurer of the claim, the better.

The insurer will typically appoint defense counsel who serves two roles--both intervention and/or stealth repair, taking a collaborative approach with the insured to fix the issue. As counsel for professional liability claims, defense counsel has typically seen hundreds, if not thousands of mistakes--all with varying degrees of consequences. They know how to identify and correct mistakes before they escalate, safeguarding the insured's interests.

First, defense counsel will assess the mistake, advise the insured on notifying their client under Rules of Professional Conduct, Rule 1.4, and determine whether the representation can ethically continue, including conflict assessment, disclosure of the error to the client, discussion of implications, obtaining written informed consent, referral to independent counsel and ongoing conflict monitoring.

If the insured does not want to do the foregoing, defense counsel will advise that they must withdraw from the representation. Then, defense counsel will assess the options to control the ramifications of the error--whether the mistake is reparable, identifying the scope of the injury and potential claimants, retention of new or additional counsel for the client (i.e., appellate counsel or practice area specialist), and potential early mediation of the dispute.

If the claim cannot be settled early and proceeds to litigation, defense counsel will zealously defend the insured against the claim (now a lawsuit) until it resolves or ends in judgment.

Mistakes happen; what counts is the response. Check AI's work early, and if something slips, get defense counsel involved fast.

Randall A. Miller is founding partner, and Jeanette Chu is an associate at Miller Waxler LLP.