A recent Law 360 story by Sarah Martinson, “Calif. Bar Adopts AI Guidelines Tackling Confidentiality, Billing”, reports that the California bar has approved guidance for attorneys using generative software — systems that produce images and text — addressing confidentiality, competence, nonlawyer supervision and billing, as other states consider passing similar guidelines.

Erika Doherty, program director for the bar's Office of Professional Competence, said at a meeting of the California bar's board of trustees that the guidelines would be the first addressing large language models, or LLMs, to be approved by a regulatory agency for lawyers.  "The benefit of that is that there is guidance available for lawyers as of right now," she said.  "The downside of that is that this is an evolving technology, and so it's going to need to be continually updated."

The guidelines advise attorneys not to input client information into any generative tool without sufficient confidentiality and security protections, and to consult with information technology or cybersecurity professionals about whether an system meets strict confidentiality, security and data retention requirements.

To meet competency requirements, attorneys should understand how the tech, generally referred to as generative AI, works and its limitations before using it, the guidelines say.  "Overreliance on AI tools is inconsistent with the active practice of law and application of trained judgment by the lawyer," they say.  On non-lawyer supervision, the guidelines advise managerial and supervisory attorneys to establish clear policies for the use of generative AI.

The guidelines also advise attorneys to charge for actual time spent working on crafting prompts and editing outputs and not to charge for time saved using the technology.  "A fee agreement should explain the basis for all fees and costs, including those associated with the use of generative AI," they say.

Since OpenAI LLC released its chatbot ChatGPT in November 2022, lawmakers and state bar authorities have been trying to figure out the best way to regulate the technology.  In June, a New York judge sanctioned two personal injury attorneys for submitting a ChatGPT-generated brief that cited nonexistent case law, finding the lawyers "abandoned their responsibilities" to check their work, and their behavior rose to "bad faith" when they then waited weeks to come clean.

This month, the State Bar of Michigan issued an opinion saying that judges have an ethical obligation to understand advancing technology, including so-called artificial intelligence, and to ensure its use in the legal system is consistent with the law.  And a Florida bar ethics committee released proposed guidelines for how attorneys can ethically use generative AI in their legal work.  Those guidelines addressed confidentiality, oversight, billing and advertising. Florida bar members have until January to comment on the proposed guidelines.