By Suzanne Burke Spencer
On November 23, 2023, the State Bar of California approved guidelines for generative AI use developed by the State Bar’s Committee on Professional Responsibility and Conduct. The “Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law” (“Guidance”) is based on lawyers’ current professional responsibility obligations, including as set forth in the California Rules of Professional Conduct and State Bar Act (Cal. Bus. & Prof. Code §§ 6000-6043). The Guidance addresses issues arising from the use of generative AI and products in which generative AI is a part. These products range from those freely available to custom-built proprietary applications. The manner in which a lawyer’s professional obligations will apply to the use of generative AI depends on many factors, including the type of product being used, the client, matter, practice area and firm size.
The Guidance covers key areas in which lawyer professional obligations may be implicated by generative AI use. Those obligations include the duties of confidentiality, competence and diligence, to comply with the law and to supervise lawyers and nonlawyers. Generative AI use may also implicate the duties of subordinate lawyers and the duties of client communication, candor to the tribunal and with respect to attorney fees and client engagement agreements. The ethical prohibition on discrimination, harassment and retaliation may also be implicated, as well as the rules of other jurisdictions in which the lawyer may be admitted to practice.
Confidentiality (rules 1.6 and 1.8.2 and Bus. & Prof. Code § 6068(e))
Lawyers are prohibited from using or disclosing client confidential information without the client’s informed consent. Generative AI may use information provided to the AI tool, including queries or uploaded documents, to train the AI and may share queries with third parties. The Guidance cautions that lawyers “must not input any confidential information of the client into any generative AI solution that lacks adequate confidentiality and security protections.” The Guidance does not specify what those adequate protections must be, but cautions lawyers to consult with IT professionals and cybersecurity experts to ensure that any AI platform used “adheres to stringent security, confidentiality, and data retention protocols.”
The Guidance further provides that lawyers must “anonymize client information and avoid entering details that can be used to identify the client.” This prohibition would again appear to apply only where the client has not given informed consent to do so. It also seems more applicable to the use of generative AI that is freely available, as opposed to a custom-built proprietary system.
Competence and Diligence Rules
Competence and diligence (rules 1.1 and 1.3)
A lawyer’s duty of competence includes the duty to keep abreast of the benefits and risks of relevant technology. To that end, the Guidance cautions that, before using generative AI, lawyers “should understand to a reasonable degree how the technology works [and] its limitations.” Even with that understanding, lawyers must not overly rely on AI tools and must instead exercise independent professional judgment and critical analysis. The Guidance therefore provides that while AI-generated outputs may be used as a “starting point,” they must be scrutinized and “critically analyzed for accuracy and bias, supplemented, and improved, if necessary.”
Because a lawyer’s professional judgment cannot be delegated to generative AI, lawyers must do more than simply detect and eliminate false AI-generated results to fulfill their professional responsibilities. Critical analysis of AI output, including ensuring the client’s interest and priorities are furthered by its use, is essential.
Other Ethical Rules Implicated by Use of AI
Compliance with the law (rules 1.2.1 and 8.4 and Bus. & Prof. Code § 6068(a))
Lawyers may not counsel a client to violate or assist a client in violating a law, rule or order of a tribunal. To ensure compliance with this ethical mandate, the Guidance cautions lawyers when using generative AI tools to “analyze the relevant laws and regulations applicable to the attorney or the client,” which may include laws specific to AI, privacy, cross-border data transfer, intellectual property and cybersecurity.
Supervisory and subordinate lawyers (rules 5.1, 5.2 and 5.3)
Firm-wide policies regarding permissible uses of generative AI should be established by managerial and supervisory lawyers and include training and adoption of reasonable measures to give reasonable assurance that the ethical and practical aspects of generative AI use, including its pitfalls, are understood by the firm’s lawyers. Subordinate lawyers may not use generative AI in a manner that violates professional obligations, even if directed by a supervisory lawyer to do so.
Client communication (rules 1.2 and 1.4)
Lawyers should consider disclosing to clients the intention to use generative AI in the client’s representation. Whether to disclose may include consideration of the novelty of the technology, the risks associated with its use, the scope of representation and client sophistication. Client disclosure may include the manner in which the technology will be used and its risks and benefits. Client instructions or guidelines restricting or limiting generative AI use must also be followed.
Candor to the tribunal (rules 3.1 and 3.3)
To be consistent with a lawyer’s duty of candor to the court, AI-generated citations to authority must be analyzed carefully for accuracy, and any inaccuracies or misstatements made to the court must be promptly corrected. Lawyers must also comply with any local court rules or orders requiring disclosure of generative AI use.
Attorney fees and engagement agreements (rule 1.5 and Bus. & Prof. Code §§ 6147-48)
The use of generative AI may allow lawyers to more efficiently create work product and therefore reduce the number of hours a lawyer may have otherwise spent; however, lawyers must not charge clients for the time saved through the use of generative AI. In other words, if a lawyer could prepare a document in 10 hours without using generative AI, but prepares it in 2 hours by using generative AI, the lawyer may not charge the client for 10 hours of time. Costs incurred with generative AI use may be passed on to the client “in compliance with applicable law,” which would generally mean the costs to be charged must be reasonable and disclosed to the client in the fee agreement or a similar writing. The basis for fees and costs to be incurred with generative AI use should also be explained in the fee agreement.
Prohibition on discrimination (rule 8.4.1)
Generative AI may be trained on biased information and lawyers using generative AI should be aware of this risk. Lawyers may not engage in discriminatory conduct and therefore should put in place policies and mechanisms to identify, report and address AI biases.
Other jurisdictions (rule 8.5)
California lawyers may be disciplined in California regardless of where their conduct occurred. Thus, California lawyers practicing in other jurisdictions should be mindful of the laws and regulations in those jurisdictions that may apply to generative AI use.
Suzanne Burke Spencer is the Chair of the California Lawyers Association Ethics Committee, a Certified Specialist in Legal Malpractice Law (State Bar of California Board of Legal Specialization) and the Managing Shareholder of Sall Spencer Callas & Krueger in Laguna Beach, California. The views here expressed are her own.
 All references to rules are to the California Rules of Professional Conduct, unless otherwise stated.
 Although not expressly stated in the Guidance, this prohibition presumably would apply only if the client does not give informed consent to do so.