Privilege Considerations When Using Generative Artificial Intelligence in Legal Practice

Recent advances in technology have transformed the ways in which lawyers deliver legal services, allowing for more efficient and sophisticated support to both corporate and individual clients. Among these developments, the rise of generative artificial intelligence (“GenAI”) stands out as a particularly disruptive force, challenging the long-standing norms surrounding how legal services are performed and safeguarded.

As awareness and adoption of GenAI systems continue to grow, attorneys are increasingly leveraging this technology to assist with a wide range of tasks, including legal research, contract review, due diligence, document review, and drafting correspondence, contracts, briefs, and other legal documents. While GenAI holds the potential to significantly enhance the quality and efficiency of legal services, its use in client matters also raises important concerns related to client confidentiality and protection of attorney-client privilege.

The following discussion provides an overview of attorney-client privilege issues arising from the use of GenAI in legal practice and highlights practical guidance from the American Bar Association (“ABA”) and various state bars designed to help attorneys comply with their ethical obligations to safeguard this critical protection and other confidential client information. 

Attorney-Client Privilege in the Era of GenAI: Understanding Potential Risks

The attorney-client privilege protects against disclosure of communications between a client and attorney made for the purposes of securing legal advice, so long as those communications are intended to be and are in fact kept confidential.

Attorney-client privilege is distinct from the duty of confidentiality. Although both serve to protect information shared between an attorney and a client, the privilege is narrower in scope, applying only to confidential communications made for the purpose of seeking or providing legal advice. In contrast, the duty of confidentiality is broader and prohibits attorneys from disclosing any information relating to the representation, regardless of its source.

As GenAI becomes more prevalent in legal practice, attorneys and clients must proceed with caution to avoid jeopardizing this privilege through inadvertent disclosure or misuse. For example, if an attorney uses a public GenAI system, such as ChatGPT, without proper safeguards, to assist with work relating to a client matter—or if a client independently inputs confidential information into such a platform—there is a risk that information may be exposed to third parties.¹ Depending on the tool’s terms of use and data retention practices, this disclosure could result in a waiver of privilege.²

Below are several ways in which the use of GenAI may place attorney-client privilege at risk.

• Disclosure to Third Parties. Inputting confidential client information into public GenAI platforms (e.g., ChatGPT without an enterprise agreement) may constitute disclosure to a third party.³ This can result in waiver of the attorney-client privilege, particularly if the information is not adequately protected or the terms of service allow the provider to retain or use the data.⁴
• Inadequate Understanding of the System’s Operations. Lawyers must understand the capabilities and limitations of any GenAI system they use. A lack of awareness around how a system stores or processes data may lead to unintentional waiver of privilege or ethical violations. Many GenAI tools store user inputs for model training or quality improvement.⁵ Without disabling these default settings or using paid/enterprise versions with stronger privacy protections, attorneys risk involuntarily exposing privileged content.⁶
• Client Misuse. Privilege can also be compromised if clients independently input privileged information into GenAI tools without understanding the implications of doing so, again potentially waiving confidentiality protections.⁷
• Lack of Client Consent or Disclosure. If a lawyer uses GenAI in connection with a matter without informing the client or obtaining informed consent, particularly where sensitive or identifiable information is involved, it may violate ethical obligations while increasing the risk of privilege loss.⁸
• Residual Data and Reuse. As set forth above, public GenAI systems may retain inputs as part of their training data unless otherwise restricted.⁹ This raises the possibility that privileged content could resurface in future responses to other users, undermining the confidentiality required for privilege to apply.¹⁰ Accordingly, lawyers should be careful about what prompts are entered into GenAI tools.

Ethical and Professional Responsibility Considerations

ABA Formal Opinion 512 (the “Opinion”) identifies key ethical concerns associated with the use of GenAI and offers general guidance to help attorneys navigate this emerging landscape. As discussed in the Opinion, several of the ABA Model Rules of Professional Conduct have implications for AI use; however, those set forth below are among the most critical when evaluating how the use of GenAI may affect an attorney’s duty to protect privileged communications and confidential client information.

Model Rule 1.1 – Competence

Model Rule 1.1 mandates lawyers provide competent representation to clients. This duty requires attorneys to exercise the “legal knowledge, skill, thoroughness, and preparation reasonably necessary for the representation” and understand “the benefits and risks associated” with the technologies used to deliver legal services to clients.¹¹

As outlined in the Opinion, in order to competently use a GenAI system in a client representation, lawyers must have a reasonable understanding of the capabilities and limitations of the specific GenAI technology at issue. This includes either acquiring a reasonable understanding of the risks and benefits of the GenAI tool employed or consulting with individuals who have the requisite technical proficiency.  The Standing Committee (the “Committee”) appreciates the rapid evolution of GenAI technologies and encourages lawyers to stay informed by reading publications about GenAI systems targeted at the legal profession, attending relevant continuing legal education programs, and consulting others who are proficient in GenAI technology.

Model Rule 1.6 – Confidentiality of Information

Model Rule 1.6 requires lawyers using GenAI to keep all information relating to the representation of a client confidential, regardless of its source, unless the client gives informed consent, disclosure is impliedly authorized to carry out the representation, or disclosure is permitted by an exception.¹²

As stated in the Opinion, when determining whether information relating to any representation is adequately protected, “lawyers must assess the likelihood of disclosure and unauthorized access, the sensitivity of the information, the difficulty of implementing safeguards, and the extent to which safeguards negatively impact the lawyer’s ability to represent the client.”¹³ Thus, before inputting information related to the representation of a client into a GenAI system, lawyers must evaluate the risks that the information will be disclosed or accessed by others both inside and outside the firm.  Notably, the Committee emphasized that GenAI systems differ in their ability to ensure information related to the representation is protected from impermissible disclosure and access. Therefore, the risk analysis will, once again, be fact-driven and depend on the client, matter, task, and GenAI system used to perform it.

The Opinion cautions lawyers against inputting confidential information into self-learning GenAI systems, which, by their nature, inherently increase the risk that information relating to one’s client representation may be disclosed improperly, even if the tool is used exclusively by lawyers at the same firm.¹⁴ However, if a lawyer chooses to do so, a client’s informed consent is required prior to inputting information relating to the representation into such a system.

Lastly, the Opinion discussed uses of self-learning GenAI systems in legal representation when client informed consent is not required—namely when a lawyer is using the tool for idea generation in a manner that does not require inputting information relating to the representation. Under such circumstances, informed consent is not necessary.¹⁵

Practical Guidance and Safeguards

The American Bar Association, along with at least a dozen state bar ethics committees, has issued guidance addressing the ethical implications of using GenAI in legal practice.¹⁶

The State Bar of California issued one of the earliest formal statements on this topic in its November 16, 2023, publication, Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law.  In its statement, the California Standing Committee sets forth guiding principles that lawyers should observe to ensure compliance with their obligations of competence and confidentiality when incorporating GenAI into their practice.¹⁷ In particular, the statement advises that lawyers:

• Must ensure competent use of technology, including the associated benefits and risks, and apply diligence and prudence with respect to facts and law;
• Should understand to a reasonable degree how the technology works, its limitations, and the applicable terms of use and other policies governing the use and exploitation of client data by the product before use of GenAI in client representation;
• Must not input any confidential information of the client into any GenAI system that lacks adequate confidentiality and security protections;
• Must anonymize client information and avoid entering details that can be used to identify the client;
• Should consult with IT professionals or cybersecurity experts to ensure that any AI system in which a lawyer would input confidential client information adheres to stringent security, confidentiality, and data retention protocols;
• Should review the Terms of Use or other information to determine how the GenAI system utilizes inputs; and
• Should ensure that the provider does not share inputted information with third parties or utilize the information for its own use in any manner, including to train or improve its product.

Furthermore, as set forth in the Opinion, the Committee concluded earlier opinions addressing technological advancements and other innovations in legal practice provide useful guidance when evaluating a lawyer’s use of a GenAI system that requires the disclosure or storage of information relating to client representation.¹⁸ In particular, the Opinion recommends lawyers should:

• Ensure that the GenAI system is configured to preserve the confidentiality and security of information, that the obligation is enforceable, and that the lawyer will be notified in the event of a breach or service of process regarding production of client information;
• Investigate the GenAI system’s reliability, security measures, and policies, including limitations on the system’s liability;
• Determine whether the case will require an attorney to disclose their GenAI practices to a client or obtain their client’s informed consent;
• Determine whether the GenAI system retains information submitted by the lawyer before and after the discontinuation of services or asserts proprietary rights to the information; and
• Understand the risk that GenAI system servers are subject to their own failures and may be an attractive target of cyber-attacks.

As GenAI tools become more integrated into legal practice, attorneys must approach their use in client matters with caution. While these technologies offer considerable benefits in terms of efficiency and productivity, they also raise risks to the confidentiality and privilege protections at the core of the attorney-client relationship. By staying informed, understanding how GenAI tools operate, and following guidance from the ABA and state bars, lawyers can responsibly leverage these tools while upholding their ethical obligations and safeguarding privileged and confidential information.

For more information on these privilege considerations or other litigation issues, contact Meghan C. Lewallen or any member of the firm’s Litigation group.

¹ Tye, Exploring the Intersections of Privacy and Generative AI: A Dive into Attorney-Client Privilege and ChatGPT, 64 JURIMETRICS J. 309, 326 (2024).

² ABA Comm. on Ethics & Prof’l Responsibility, Formal Op. 512 at 3 (2024) (hereinafter “ABA Formal Op. 512”).

³ Tye, 64 JURIMETRICS J. at 331 (2024).

⁴ ABA Formal Op. 512 at 11.

⁵ Tye, 64 JURIMETRICS J. at 311 (2024).

⁶ Id. at 330.

⁷Id. at 326.

⁸ ABA Formal Op. 512 at 7-8.

⁹ Tye, 64 JURIMETRICS J. at 311 (2024).

¹⁰ ABA Formal Op. 512 at 6.

¹¹ MODEL RULES OF PRO CONDUCT R. 1.1 (2023) (hereinafter “MODEL RULES”); ABA Form Opinion 512 at 2.

¹² MODEL RULES R. 1.6 (2023); ABA Form Opinion 512 at 2.

¹³ ABA Formal Op. 512 at 6.

¹⁴ Id.

¹⁵ Id.

¹⁶ Chrostowski, et al., Using Artificial Intelligence in Litigation, ETHICS (This includes California, Florida, New Jersey, New York, Minnesota, Kentucky, Missouri, North Carolina, Pennsylvania, Texas, West Virginia, and D.C.)

¹⁷ State Bar of California Standing Committee on Professional Responsibility, Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law, at 2 (November 16, 2023).

¹⁸ ABA Formal Op. 512 at 11.