The American Bar Association’s (ABA) Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512, attempting to provide guidance on the ethical considerations for lawyers using generative artificial intelligence (GAI) tools. While the opinion addresses important issues, it may raise as many questions as it answers regarding the use of GAI in legal practice.
The ABA’s Formal Opinion 512 addresses key ethical considerations for lawyers using GAI, including competence, confidentiality, client communication, candor to tribunals, supervision, and fee arrangements. The opinion emphasizes that lawyers must understand the capabilities and limitations of GAI tools they use and cannot abdicate their professional judgment to GAI. The opinion also warns against uncritical reliance on AI-generated content.
According to the opinion, lawyers should take several steps to understand the capabilities and limitations of GAI tools. It emphasizes that lawyers must either acquire a “reasonable understanding” of the benefits and risks of specific GAI tools they use or consult experts who can provide such guidance. The opinion suggests reading about GAI tools targeted at the legal profession, attending relevant continuing legal education programs, and staying informed about ongoing developments in GAI technology.
Importantly, the opinion advises that, at a minimum, lawyers should read and understand the terms of use, privacy policy, and related contractual terms of any GAI tool they employ. If lawyers are unable to fully grasp these terms, they should consult with colleagues or external experts who have analyzed them. The opinion also notes that lawyers may need to consult with IT professionals or cybersecurity experts to fully understand how GAI tools utilize information. However, the ABA does not provide a clear definition of what constitutes a “reasonable understanding,” potentially leaving lawyers uncertain about the extent of knowledge required to meet their ethical obligations.
Further, the opinion discusses several situations in which disclosure and informed client consent regarding use of GAI are needed. For examples, it advises that for “self-learning” GAI systems, which the opinion describes as GAI systems that “learn from themselves as they cull more data,” lawyers must obtain informed client consent before inputting any information relating to the client representation. Although the opinion does provide an example of a permissible use of “self-learning” GAI that does not require client consent (i.e., to generate ideas without inputting client-specific information), the phrase “any information relating to client representation” leaves room for interpretation.
As a result, while the opinion provides some needed guidance, its cautious approach may not fully account for the transformative potential of GAI in legal practice. However, the Committee acknowledges the rapidly evolving nature of GAI technology, conceding that “as [GAI] technology develops, the risks may change in ways that would alter our conclusion.” It is important to note that ABA formal opinions, while persuasive, are not binding and may not reflect the approach ultimately adopted by state bars or courts.[1]
Formal Opinion 512 is available here.
[1]See, e.g., Florida Bar Ethics Opinion 24-1, available at: https://www.floridabar.org/etopinions/opinion-24-1/ (recommending that a lawyer obtain the “affected client’s informed consent prior to utilizing a third-party generative AI program if the utilization would involve the disclosure of any confidential information.”); see also State Bar of Cal. Standing Comm. on Prof’l Resp. & Conduct, Practical Guidance For the Use of Generative Artifical Intelligence in the Practice of Law (2024), available at: https://www.calbar.ca.gov/Portals/0/documents/ethics/Generative-AI-Practical-Guidance.pdf (advising that “[a] lawyer must not input any confidential information of the client into any generative AI solution that lacks adequate confidentiality and security protections. A lawyer must anonymize client information and avoid entering details that can be used to identify the client.”).
