The Akron Legal News

Login | April 12, 2024

California and Florida bars promulgate GenAI guidelines

Technology for Lawyers

Published: March 8, 2024

The California Lawyers Association and the Florida Bar Association have created some AI ethics guidelines, published around the end of last year and the first of this year.
The guidelines are similar. Looks like a pretty good first stab. Of course, making rules like this is like trying to ride a comet, but here we go. As usual, California is taking the lead. But Florida, surprisingly, is right there, and both states are taking similar approaches.
Let’s take a look.
Both states identify similar potential problems with lawyers using generative (writing) AI. These include confidentiality, supervision, legal fees and costs, advertising, and general all-round familiarity with these potential problems.
Confidentiality: Lawyers have to familiarize themselves with the technology at least to the extent that they know where their data is, so that confidential data won’t be compromised inadvertently.
So we need to be clear here: GenAI gives you access to one of various supercomputers that are owned by companies that sell data.
If you pump in a client’s confidential data into one of these supercomputers, there is a more-than-zero chance that someone else will see it. And even if that peek at that data goes no further, you have compromised client confidential information.
The opinions note that the way to mitigate this problem may be with the use of in-house GenAI, but only the largest firms could even have access to that kind of technology.
Supervision: The new rules mirror a growing trend of treating GenAI like it’s a paralegal, with the same supervisory rules as that would entail.
So—if you ask a GenAI chatbot a legal question, it must respond with a disclaimer that it is not a lawyer.
Obviously, all work has to be checked by a lawyer as if it were a non-attorney helper. And this also applies to clients: lawyers should advise clients not to use GenAI in making reports or procuring something evidentiary.
I think it also goes without saying that this does not apply to AI-based discovery tools, but that’s another issue.
Legal Fees and Costs: Which is yeah don’t charge clients as if you had actually done the work that the GenAI is doing.
Like a research project would have taken eight hours if you had done it but GenAI did it in 15 seconds, you don’t charge for eight hours.
Advertising and Intake: Make sure that the front-facing chatbot tells the truth to clients and potential clients.
Imputation of Knowledge: If you’re going to use this stuff, you are bound to know how it works. Shout me out if you have any questions.
Check California’s rules out here: