Attorneys tasked with supervising “non-human, non-lawyer” assistant

CYBERSLEUTH

Maya Angelou said, "Do the best you can until you know better. Then when you know better, do better." The California State Bar is hoping that its new guidance on the ethical use of generative artificial intelligence will help attorneys do better - currently a very low bar.

This year numerous stories have emerged of attorneys blindly (and probably unethically) using generative AI to prepare legal pleadings without taking the most rudimentary steps to ensure their accuracy. The most infamous example occurred in May when a federal district court judge for the Southern District of New York issued a scathing order to show cause against two attorneys who generated a brief with artificial intelligence platform ChatGPT without checking the cases cited. Roberto Mata v. Avianca, Inc., Case No. 22-cv-1461 (PKC).

The ChatGPT brief was full of "bogus judicial decisions with bogus quotes and bogus internal citations," according to the judge. One of the attorneys told the court that he was a first-time user of ChatGPT and "was unaware of the possibility that its content could be false."

It should go without saying that attorneys must check the accuracy of any practice aids before submitting pleadings based on them, but apparently it must be said. Earlier this month, the California Bar's board of trustees approved new AI guidance, which reminds attorneys that the use of "non-human, non-lawyer" assistance is subject to the same ethical rules as any other activity.

The Bar's AI guidance laments that attorneys might be duped into a false sense of security because AI generates output "that projects confidence and effectively emulates human responses." Maybe so, but it wouldn't be acceptable to simply copy the work product of a human attorney and pass it off as one's own. The Bar AI guidance observes that "a lawyer's professional judgment cannot be delegated to generative AI and remains the lawyer's responsibility at all times."

Most attorneys are unaware that generic AI platforms do not have access to legal databases such as Lexis and Westlaw. Instead, they draw data from publicly available sources, which might include fictional legal material from Boston Legal or LA Law. Not only can AI generate bogus case cites, but the data collected by some AI platforms is a year out of date, so even real cases might have been reversed or overridden by statutes since the last data update. The Bar's AI guidance admonishes that "the duty of competence requires more than the mere detection and elimination of false AI-generated results."

These are helpful, if not completely obvious, observations. What is the Bar's solution? Beginning with reporting year 2025, all attorneys must complete one hour of continuing legal education in use of technology in the practice of law, and the California Bar's AI guidance calls for the creation of a course on use of AI that can be used to fulfill this requirement. State Bar rule 2.72(C)(2)(a)(iv). Waiting a year to offer a course on AI as part of MCLE is far too anemic of a response.

The Bar AI guidance notes that AI is "a rapidly evolving technology that presents novel issues," so the response should be commensurate. To have any chance of keeping even reasonably timely pace with AI developments, the Bar needs to provide a resource that can quickly respond to new and novel situations. The Bar should utilize the work group of industry experts on generative AI it assembled to prepare the guidance document to provide monthly blogs or practice advisories.

The Bar's AI guidance notes that use of the technology could violate an attorney's duty of confidentiality. To use AI platforms, attorneys type prompts into the system or may upload client documents specifying what they want to generate. The Bar's AI guidance notes that the queries may be shared with third parties, and thereby disclose client confidences.

The Bar recommends that attorneys review the terms of use of AI platforms to ensure that inputted information is not shared with third parties. But unless an attorney is using an AI platform internal to his or her firm or company, it is literally impossible to avoid sharing information with third parties. Generative AI learns from its own experience, and many (if not all) AI platforms record the prompts and interactions with the requesting attorney, store the transcript for an indefinite period and use the content to improve the AI program.

The Bar's AI guidance recommends (but does not require) that attorneys disclose to their client that they intend to use generative AI in their legal representation, including how the technology will be used, and the benefits and risks of such use. Such disclosure will inevitably cause clients to ask how the use of AI affects billable hours. The Bar's AI guidance offers an answer.

"A lawyer may use generative AI to more efficiently create work product and may charge for actual time spent ... [but] must not charge hourly fees for the time saved by using generative AI." That should seem obvious - attorneys can't bill for hours not actually spent working whether it's because work was delegated to AI platforms or because the attorney is billing for the "value" of his or her work regardless of the actual hours spent.

AI's greatest contribution may be that it has laid bare questionable legal practices that violate an attorney's ethical obligations regardless of technology. As the Bar's AI guidance notes, all of the issues arising from generative AI so far fall squarely within the existing rules of professional responsibility.