Proposed Advisory Opinion 24-1 Regarding Lawyers’ use of Generative Artificial Intelligence – Official Notice

Proposed Advisory Opinion 24-1 Regarding Lawyers’ use of Generative Artificial Intelligence – Official Notice

The Florida Bar Board of Governors’ Review Committee on Professional Ethics has issued Proposed Advisory Opinion 24-1, reprinted below. Pursuant to Rule 6(d) and (e) of The Florida Bar Procedures for Ruling on Questions of Ethics, comments from Florida Bar members are solicited on the proposed opinion. The board will consider any comments received at a meeting scheduled to be held on Friday, January 19, 2024, at the AC Hotel in Tallahassee, Florida. Comments must contain the proposed advisory opinion number and clearly state the issues for the committee to consider. A written argument may be included explaining why the Florida Bar member believes the committee’s opinion is either correct or incorrect and may contain citations to relevant authorities. Comments should be submitted to Jonathan D. Grabb, Ethics Counsel, The Florida Bar, 651 E. Jefferson Street, Tallahassee 32399-2300, or emailed to [email protected], and must be postmarked no later than January 2, 2024.

PROFESSIONAL ETHICS OF THE FLORIDA BAR

Proposed Advisory Opinion 24-1

[DATE]

The Florida Bar Board of Governors has directed the Board Review Committee on Professional Ethics to issue an opinion regarding lawyers’ use of generative artificial intelligence (“AI”). The release of ChatGPT-3 in November 2022 prompted wide-ranging debates regarding lawyers’ use of generative AI in the practice of law. While it is impossible to determine the impact generative AI will have on the legal profession, this opinion is intended to provide guidance to Florida Bar members regarding some of the ethical implications of these new programs.

Generative AI are “deep-learning models” that compile data “to generate statistically probable outputs when prompted.” IBM, What is generative AI?, (April 20, 2023), https://research.ibm.com/blog/what-is-generative-AI (last visited 11/09/2023). Generative AI can create original images, analyze documents, and draft briefs based on written prompts. Often, these programs rely on large language models. The datasets utilized by generative AI large language models can include billions of parameters making it virtually impossible to determine how a program came to a specific result. Tsedel Neeley, 8 Questions About Using AI Responsibly, Answered, Harv. Bus. Rev. (May 9, 2023).

While generative AI may have the potential to dramatically improve the efficiency of a lawyer’s practice, it can also pose a variety of ethical concerns. Among other pitfalls, lawyers are quickly learning that generative AI can “hallucinate” or create “inaccurate answers that sound convincing.” Matt Reynolds, vLex releases new generative AI legal assistant, A.B.A. J. (Oct. 17, 2023), https://www.abajournal.com/web/article/vlex-releases-new-generative-ai-legal-assistant (last visited 11/09/2023). In one particular incident, a federal judge sanctioned two unwary lawyers and their law firm following their use of false citations created by generative AI. Mata v. Avianca, 22-cv-1461, 2023 WL 4114965, at 17 (S.D.N.Y. June 22, 2023).

Even so, the judge’s opinion explicitly acknowledges that “[t]echnological advances are commonplace and there is nothing inherently improper about using a reliable artificial intelligence tool for assistance.” Id. at 1.

Due to these concerns, lawyers using generative AI must take reasonable precautions to protect the confidentiality of client information, develop policies for the reasonable oversight of generative AI use, ensure fees and costs are reasonable, and comply with applicable ethics and advertising regulations.

Confidentiality

A lawyer’s first responsibility when using generative AI should be the protection of the confidentiality of the client’s information as required by Rule 4-1.6 of the Rules Regulating The Florida Bar. The ethical duty of confidentiality is broad in its scope and applies to all information learned during a client’s representation, regardless of its source. Rule 4-1.6, Comment. Absent the client’s informed consent or an exception permitting disclosure, a lawyer may not reveal the information. In practice, the most common exception is found in subdivision (c)(1), which permits disclosure to the extent reasonably necessary to “serve the client’s interest unless it is information the client specifically requires not to be disclosed[.]” Rule 4-1.6(c)(1). Nonetheless, it is recommended 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.

Rule 4-1.6(e) also requires a lawyer to “make reasonable efforts to prevent the inadvertent or unauthorized disclosure of, or unauthorized access to, information relating to the client’s representation.” Further, a lawyer’s duty of competence requires “an understanding of the benefits and risks associated with the use of technology[.]” Rule 4-1.1, Comment.

When using a third-party generative AI program, lawyers must sufficiently understand the technology to satisfy their ethical obligations. For generative AI, this specifically includes knowledge of whether the program is “self-learning.” A generative AI that is “self-learning” continues to develop its responses as it receives additional inputs and adds those inputs to its existing parameters. Neeley, supra n. 2. Use of a “self-learning” generative AI raises the possibility that a client’s information may be stored within the program and revealed in response to future inquiries by third parties.

Existing ethics opinions relating to cloud computing, electronic storage disposal, remote paralegal services, and metadata have addressed the duties of confidentiality and competence to prior technological innovations and are particularly instructive. In its discussion of cloud computing resources, Florida Ethics Opinion 12-3 cites to New York State Bar Ethics Opinion 842 and Iowa Ethics Opinion 11-01 to conclude that a lawyer should:

• Ensure that the provider has an obligation to preserve the confidentiality and security of information, that the obligation is enforceable, and that the provider will notify the lawyer in the event of a breach or service of process requiring the production of client information;
• Investigate the provider’s reputation, security measures, and policies, including any limitations on the provider’s liability; and
• Determine whether the provider retains information submitted by the lawyer before and after the discontinuation of services or asserts proprietary rights to the information.

While the opinions were developed to address cloud computing, these recommendations are equally applicable to a lawyer’s use of third-party generative AI when dealing with confidential information.

Florida Ethics Opinion 10-2 discusses the maintenance and disposition of electronic devices that contain storage media and provides that a lawyer’s duties extend from the lawyer’s initial receipt of the device through the device’s disposition, “including after it leaves the control of the lawyer.” Opinion 10-2 goes on to reference a lawyer’s duty of supervision and to express that this duty “extends not only to the lawyer’s own employees but over entities outside the lawyer’s firm with whom the lawyer contracts[.]” Id.

Florida Ethics Opinion 07-2 notes that a lawyer should only allow an overseas paralegal provider access to “information necessary to complete the work for the particular client” and “should provide no access to information about other clients of the firm.” Additionally, while “[t]he requirement for informed consent from a client should be generally commensurate with the degree of risk involved[,]” including “whether a client would reasonably expect the lawyer or law firm to personally handle the matter and whether the non-lawyers will have more than a limited role in the provision of the services.” Id. Again, this guidance seems equally applicable to a lawyer’s use of generative AI.

Finally, Florida Ethics Opinion 06-2 provides that a lawyer should take reasonable steps to safeguard the confidentiality of electronic communications, including the metadata attached to those communications, and that the recipient should not attempt to obtain metadata information that they know or reasonably should know is not intended for the recipient. In the event that the recipient inadvertently receives metadata information, the recipient must “promptly notify the sender,” as is required by Rule 4-4.4(b). Similarly, a lawyer using generative AI should take reasonable precautions to avoid the inadvertent disclosure of confidential information and should not attempt to access information previously provided to the generative AI by other lawyers.

It should be noted that confidentiality concerns may be mitigated by use of an inhouse generative AI rather than an outside generative AI where the data is hosted and stored by a third-party. If the use of a generative AI program does not involve the disclosure of confidential information to a third-party, a lawyer is not required to obtain a client’s informed consent pursuant to Rule 4-1.6.

Oversight of Generative AI

While Rule 4-5.3(a) defines a nonlawyer assistant as a “a person,” many of the standards applicable to nonlawyer assistants provide useful guidance for a lawyer’s use of generative AI.

First, just as a lawyer must make reasonable efforts to ensure that a law firm has policies to reasonably assure that the conduct of a nonlawyer assistant is compatible with the lawyer’s own professional obligations, a lawyer must do the same for generative AI. Lawyers who rely on generative AI for research, drafting, communication, and client intake risk many of the same perils as those who have relied on inexperienced or overconfident nonlawyer assistants.

Second, a lawyer must always review the work product of a generative AI just as the lawyer must do so for the work of nonlawyer assistants such as paralegals. Lawyers are ultimately responsible for the work product that they create regardless of whether that work product was originally drafted or researched by a nonlawyer or generative AI.

Functionally, this means a lawyer must verify the accuracy and sufficiency of all research performed by generative AI. The failure to do so can lead to violations of the lawyer’s duties of competence (Rule 4-1.1), avoidance of frivolous claims and contentions (Rule 4-3.1), candor to the tribunal (Rule 4-3.3), and truthfulness to others (Rule 4-4.1), in addition to sanctions that may be imposed by a tribunal against the lawyer and the lawyer’s client.

Third, these duties apply to nonlawyers “both within and outside of the law firm.” ABA Comm. on Ethics and Prof’l Responsibility, Formal Op. 498 (2021); see Fla. Ethics Op. 07-2. The fact that a generative AI is managed and operated by a third-party does not obviate the need to ensure that its actions are consistent with the lawyer’s own professional and ethical obligations.

Further, a lawyer should carefully consider what functions may ethically be delegated to generative AI. Existing ethics opinions have identified tasks that a lawyer may or may not delegate to nonlawyer assistants and are instructive. First and foremost, a lawyer may not delegate to generative AI any act that could constitute the practice of law such as the negotiation of claims or any other function that requires a lawyer’s personal judgment and participation.

Florida Ethics Opinion 88-6 notes that, while nonlawyers may conduct the initial interview with a prospective client, they must:

• Clearly identify their nonlawyer status to the prospective client;
• Limit questions to the purpose of obtaining factual information from the prospective client; and
• Not offer any legal advice concerning the prospective client’s matter or the representation agreement and refer any legal questions back to the lawyer.

This guidance is especially useful as law firms increasingly utilize website chatbots for client intake. While generative AI may make these interactions seem more personable, it presents additional risks, including that a prospective client relationship or even a lawyer-client relationship has been created without the lawyer’s knowledge.

The Comment to Rule 4-1.18 (Duties to Prospective Client) explains what constitutes a consultation:

A person becomes a prospective client by consulting with a lawyer about the possibility of forming a client-lawyer relationship with respect to a matter. Whether communications, including written, oral, or electronic communications, constitute a consultation depends on the circumstances. For example, a consultation is likely to have occurred if a lawyer, either in person or through the lawyer’s advertising in any medium, specifically requests or invites the submission of information about a potential representation without clear and reasonably understandable warnings and cautionary statements that limit the lawyer’s obligations, and a person provides information in response. In contrast, a consultation does not occur if a person provides information to a lawyer in response to advertising that merely describes the lawyer’s education, experience, areas of practice, and contact information, or provides legal information of general interest. A person who communicates information unilaterally to a lawyer, without any reasonable expectation that the lawyer is willing to discuss the possibility of forming a client-lawyer relationship, is not a “prospective client” within the meaning of subdivision (a).

Similarly, the existence of a lawyer-client relationship traditionally depends on the subjective reasonable belief of the client regardless of the lawyer’s intent. Bartholomew v. Bartholomew, 611 So. 2d 85, 86 (Fla. 2d DCA 1992).

For these reasons, a lawyer should be wary of utilizing an overly welcoming generative AI chatbot that may provide legal advice, fail to immediately identify itself as a chatbot, or fail to include clear and reasonably understandable disclaimers limiting the lawyer’s obligations.

Just as with nonlawyer staff, a lawyer should not instruct or encourage a client to rely solely on the “work product” of generative AI, such as due diligence reports, without the lawyer’s own personal review of that work product.

Legal Fees and Costs

Rule 4-1.5(a) prohibits lawyers from charging, collecting, or agreeing to fees or costs that are illegal or clearly excessive while subdivision (b) provides a list of factors to consider when determining whether a fee or cost is reasonable. A lawyer must communicate the basis for fees and costs to a client and it is preferable that the lawyer do so in writing. Rule 4-1.5(e). Contingent fees and fees that are nonrefundable in any part must be explained in writing. Rule 4-1.5(e); Rule 4-1.5(f)(2).

Regarding costs, a lawyer may only ethically charge a client for the actual costs incurred on the individual client’s behalf and must not duplicate charges that are already accounted for in the lawyer’s overhead. See The Florida Bar v. Carlon, 820 So. 2d 891, 899 (Fla. 2002) (lawyer sanctioned for violations including a $500.00 flat administrative charge to each client’s file); ABA Comm. on Ethics and Prof’l Responsibility, Formal Op. 93-379 (1993) (lawyer should only charge clients for costs that reasonably reflect the lawyer’s actual costs); Rule 4-1.5(h) (lawyers accepting payment via a credit plan may only charge the actual cost imposed on the transaction by the credit plan).

Regarding fees, a lawyer may not ethically engage in any billing practices that duplicate charges or that falsely inflate the lawyer’s billable hours. Though generative AI programs may make a lawyer’s work more efficient, this increase in efficiency must not result in falsely inflated claims of time. In the alternative, lawyers may want to consider adopting contingent fee arrangements or flat billing rates for specific services so that the benefits of increased efficiency accrue to the lawyer and client alike.

While a lawyer may separately itemize activities like paralegal research performed by nonlawyer personnel, the lawyer should not do so if those charges are already accounted for in the lawyer’s overhead. Fla. Ethics Op. 76-33 & 76-38, Consolidated. In the alternative, the lawyer may need to consider crediting the nonlawyer time against the lawyer’s own fees. Id. Florida Ethics Opinion 07-2 discusses the outsourcing of paralegal services in contingent fee matters and explains:

The law firm may charge a client the actual cost of the overseas provider [of paralegal services], unless the charge would normally be covered as overhead. However, in a contingent fee case, it would be improper to charge separately for work that is usually otherwise accomplished by a client’s own attorney and incorporated into the standard fee paid to the attorney, even if that cost is paid to a third party provider.

Additionally, a lawyer should have sufficient general knowledge to be capable of providing competent representation. See, e.g., Att’y Grievance Comm’n of Maryland v. Manger, 913 A.2d 1 (Md. 2006). “While it may be appropriate to charge a client for case-specific research or familiarization with a unique issue involved in a case, general education or background research should not be charged to the client.” Id. at 5.

In the context of generative AI, these standards require a lawyer to inform a client, preferably in writing, of the lawyer’s intent to charge a client the actual cost of using generative AI. In all instances, the lawyer must ensure that the charges are reasonable and are not duplicative. If a lawyer is unable to determine the actual cost associated with a particular client’s matter, the lawyer may not ethically prorate the periodic charges of the generative AI and instead should account for those charges as overhead. Finally, while a lawyer may charge a client for the reasonable time spent for case-specific research and drafting when using generative AI, the lawyer should be careful not to charge for the time spent developing minimal competence in the use of generative AI.

Lawyer Advertising

The advertising rules in Subchapter 4-7 of the Rules Regulating The Florida Bar include prohibitions on misleading content and unduly manipulative or intrusive advertisements.

Rule 4-7.13 prohibits a lawyer from engaging in advertising that is deceptive or inherently misleading. More specifically, subdivision (b) includes prohibitions on:

(3) comparisons of lawyers or statements, words, or phrases that characterize a lawyer’s or law firm’s skills, experience, reputation, or record, unless the characterization is objectively verifiable; [and]

….

(5) [use of] a voice or image that creates the erroneous impression that the person speaking or shown is the advertising lawyer or a lawyer or employee of the advertising firm unless the advertisement contains a clear and conspicuous disclaimer that the person is not an employee or member of the law firm[.]

As noted above, a lawyer should be careful when using a generative AI chatbot for advertising and intake purposes as the lawyer will be ultimately responsible in the event the chatbot provides misleading information to prospective clients or communicates in a manner that is inappropriately intrusive or coercive. To avoid confusion, a lawyer should inform prospective clients that they are communicating with an AI program and not with a lawyer or law firm employee. Additionally, while many visitors to a lawyer’s website voluntarily seek information regarding the lawyer’s services, a lawyer should consider including screening questions that limit the chatbot’s communications if a person is already represented by another lawyer.

Lawyers may advertise their use of generative AI but cannot claim their generative AI is superior to those used by other lawyers or law firms unless the lawyer’s claims are objectively verifiable. Whether a particular claim is capable of objective verification is a factual question that must be made on a case-by-case basis.

Conclusion

In sum, a lawyer may ethically utilize generative AI technologies but only to the extent that the lawyer can reasonably guarantee compliance with the lawyer’s ethical obligations. These obligations include the duties of confidentiality, avoidance of frivolous claims and contentions, candor to the tribunal, truthfulness in statements to others, avoidance of clearly excessive fees and costs, and compliance with restrictions on advertising for legal services. Lawyers should be cognizant that generative AI is still in its infancy and that these ethical concerns should not be treated as an exhaustive list. Rather, lawyers should continue to develop competency in their use of new technologies and the risks and benefits inherent in those technologies.