On July 29, 2024, the American Bar Association’s Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512, entitled “Generative Artificial Intelligence Tools.” This opinion joins good ones from Florida and California in providing helpful guidance to lawyers on how to ethically incorporate generative AI—a subset of AI technology—into your practice.1 Opinion 512 is recommended to all who have even a minor interest in this topic, have considered using generative AI tools, or have already been using such tools in their practice. Because Minnesota generally follows the model rules of relevance on this topic, Opinion 512 is particularly instructive for Minnesota lawyers. This column presents a high-level summary.
The opinion starts with a good reminder to us all: Artificial intelligence tools have long been used in legal practice—electronic discovery, data analytics, and legal research, to name a few. Thus, lawyers are or should be familiar generally with how to ethically incorporate technology tools into their legal practice. Generative AI takes those technologies further by creating new content from large data sets of information, and continues to evolve in scope and use. But understanding the ethical implications of generative AI follows the same path as one would take toward the incorporation of any new technology tool into your practice, and is a good reminder that we should be analyzing all technology and other third-party or vendor support through the same lens.
Ethics issues to consider
The opinion summary hits the ethics duties implicated: “To ensure clients are protected, lawyers using generative artificial intelligence tools must fully consider their applicable ethics obligations, including their duties to provide competent legal representation, to protect client information, to communicate with clients, to supervise their employees and agents, to advance only meritorious claims and contentions, to ensure candor toward the tribunal, and to charge reasonable fees.”
The opinion starts with your duty of competence. Of note, the opinion does not expect us to be experts in the technology to be able to use it in our practice, but does say we must have a “reasonable understanding of the capabilities and limitations” before doing so. Self-study could satisfy this obligation, but remember that the technology is rapidly advancing, so this is not a one-and-done task.
The opinion next takes on confidentiality, a topic that has been much discussed with generative AI since many forms retain inputs as part of its learning, and thus could lead to unauthorized disclosure of confidential client information. In particular, the opinion covers some ways that information can be improperly disclosed even when using in-house-only generative tools, a topic that I admit I had not previously given much thought. The opinion posits, and I tend to agree, that the use of many generative AI tools (beyond simple idea-generation tools) will generally require a client’s informed consent. And remember, you have to explain information with particularity in order to be able to characterize a client’s consent as “informed.”2 Generic, boiler-plate language in engagement agreements is an insufficient way to obtain informed consent.
Your communication obligation comes into play as well with generative AI tools. Clients may want to know (and thus you must be able to explain) your use of generative AI in your practice, but sometimes you must communicate your use unprompted even when informed consent is not required. One example provided in the opinion is when a lawyer uses generative AI to evaluate and advise on jury selection, as a client would reasonably expect to be advised of how much the lawyer is deferring to generative AI outputs versus the lawyer’s own independent judgment. In general, the opinion recommends explaining to clients how you use generative AI tools to assist in your delivery of legal services as part of effective client communications.
The opinion next covers risks of generative AI use that lead to issues of candor to tribunals (such as case “hallucinations,” or arguments without merit). You should review all output that is going to be incorporated into work product presented to a tribunal for accuracy, just as you would any other sources you cite. Further, we can rely on the work of others, but we must take steps to ensure it is accurate. Failure to do so may implicate several rules.
The opinion also covers the duty of supervision. Whether generative AI use is permitted in your workplace should be the subject of a firm or office policy. Training should be provided if generative AI tools are used, and you should understand if outside vendors or service providers are employing generative AI tools; if so, you must make efforts to ensure they are only doing so in a manner that is consistent with your ethical obligations.
And finally, the opinion covers the important but often overlooked issue of fees. If clients are paying for particular tools or services, including generative AI tools, that must be explained at the beginning of the representation, preferably in writing. If your use of generative AI makes you more efficient, your hourly billing—if you are billing hourly—should reflect that efficiency, as you can never bill for more time than actually spent. In this regard, the opinion cites to one of my favorite ABA opinions—Opinion 93-379, an oldie but still relevant regarding billing practices.3 The opinion reminds us, “Lawyers must remember that they may not charge clients for time necessitated by their own inexperience.”
Conclusion
Your duty of competence includes an obligation to understand the benefits and risks of any technology you use in your legal practice. Generative AI is more sophisticated and varied in its applications than most technology we use, and therefore requires that we take the time to carefully assess its compatibility with our ethical obligations. On our ethics hotline, it has been exciting to hear how lawyers are exploring specific products to better serve clients and maximize available resources, and it has been interesting to help them through the various rules as they relate to a particular use. Opinion 512 is an additional reference to help you in this task. Remember also that ABA ethics opinions are free to all within one year of publication, but thereafter you will have to pay for the opinion if you are not an ABA member. Be sure to download ABA Opinion 512 now if you are exploring generative AI use in your practice.
Susan Humiston is the director of the Office of Lawyers Professional Responsibility and Client Security Board. Prior to her appointment, Susan worked in-house at a publicly traded company, and in private practice as a litigation attorney.
Notes
1 Florida Bar Ethics Opinion 24-1 (1/19/2024), and California’s “Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law,” both of which are available through a Google search, are also good references for Minnesota lawyers, although lawyers should be wary of modest variations between the applicable Florida, California, and Minnesota ethics rules.
2 Rule 1.0(f), Minnesota Rules of Professional Conduct (MRPC), defines “informed consent” as “the agreement by a person to a proposed course of conduct after the lawyer has communicated adequate information and explanation about the material risks of and reasonably available alternatives to the proposed course of conduct.”
3 ABA Formal Ethics Opinion 93-379 “Billing for Professional Fees, Disbursements and Other Expenses” (12/6/1993).