🔒The risks of AI for law firms and their business clients

By now, many people have heard about the New York lawyer who got into trouble earlier this year when he tried to use ChatGPT for legal research in preparing a memorandum of law on behalf of his client.

The problem for the lawyer wasn’t simply that he used a generative artificial intelligence tool in doing work for a client.

The problem was that when he was given fair warning by both opposing counsel and the court that his research may have been inadequate, he more or less doubled down on his mistakes, apparently unwilling to accept or acknowledge that he had used new technology without understanding how it worked.

Then he compounded that mistake by attempting to mislead the court in a few different ways, which is never a good idea.

Though the focus of the case was on lawyer conduct, some of its lessons apply to any profession. Don’t use new technology unless you have some notion of its limitations. And don’t think a mistake will go away if you just pretend there was no mistake.

But the case also serves as a reminder that with generative AI, a relatively new and still-developing technology, there may be risks that dictate some caution before clients can expect it to be widely used in law firms.

What are some of those risks?

1. Generative AI searches may lead to inaccurate and unreliable results.

For example, as of May 2023, the terms of use for one generative AI platform included this disclaimer: “Use of services may in some situations result in incorrect output that does not accurately reflect real people, place or facts,” and the tool “may occasionally generate incorrect information.”

Accordingly, even if the response to a generative AI query can provide a starting place for research, or provide ideas for the drafting of certain kinds of litigation or transactional documents, the proper use of any such results will require application of the type of lawyerly analysis and judgment that clients commonly expect from their lawyers.

Any lawyers using generative AI for research projects or brief writing still need to confirm the reliability of the results they obtain. Generative AI may develop to the stage where it can provide the requisite analysis, but it is not there yet.

2. A second concern is confidentiality.

Lawyers and law firms have an obligation to keep client information confidential. It is a fundamental premise of the attorney-client relationship.

But entering client-related information into a generative AI platform immediately places that information outside of the firm’s secure systems and onto third-party servers where the firm can no longer ensure that appropriate safeguards are in place to protect the information from disclosure to, and use by, others.

3. A third concern is infringement.

The content generative AI creates may infringe on the ownership rights of those whose intellectual property was used to train the platform.

And generative AI creates its responses without specific attribution to the source, or sources, on which it relies, even when those responses repeat the source material verbatim.

Lawyers use such responses at some risk of unknowingly engaging in plagiarism.

Businesses that regularly engage law firms should not assume that a firm’s use of generative-AI tools will lead to more efficient, and thus less costly, legal services.

There is no question that in the long term, generative AI will efficiently and effectively handle certain tasks, such as contract drafting and document review in discovery, leaving lawyers more time to focus on intellectual analysis.

But for now, because the technology is so new and developing so quickly, business owners may want to ask their law firms about how they use generative AI, and whether there are safeguards and protocols in place to ensure that the lawyers using those tools are competent to do so.

Marcy Tench Stovall and David P. Atkins are litigation attorneys in the professional liability practice at Pullman & Comley.