[author: Ismail Amin]

The rapid evolution of digital technologies has ushered in a new era for the legal profession—one characterized by both unprecedented promise and intricate new hazards. As practitioners and clients alike become more reliant on artificial intelligence, questions abound regarding the traditional boundaries of confidentiality and privilege that have long served as the bedrock of attorney-client relationships.

The accelerated adoption of machine learning (ML), large language models (LLM) and artificial intelligence (AI) has engendered significant transformation within the legal profession, offering new efficiencies but also introducing complex challenges with respect to confidentiality obligations and evidentiary attorney-client privileges. Among the most pressing of these is the effect of AI platforms on the sanctity of attorney-client communications and the attendant risk of inadvertently waiving the attorney-client privilege by clients.

We analyze the ramifications of clients’ engagement with AI platforms—including generative AI tools, cloud-based legal assistants, and related digital services—on privileged communications. It further evaluates pertinent legal doctrines and proposes measures that legal practitioners may adopt to preserve the integrity of confidential counsel.

I.Attorney-Client Privilege: Foundational Principles

The attorney-client privilege constitutes a sacrosanct cornerstone of the legal system, shielding the confidentiality of exchanges between attorneys and their clients. For purposes of this writing, we examine the privilege in California, Nevada, Texas and New York. This privilege is intended to facilitate candid and comprehensive communication, thereby enabling clients to seek legal guidance without apprehension that their disclosures might be exploited against their interests.

In California, California Evidence Code § 954 establishes the attorney-client privilege, allowing clients to refuse disclosure of confidential communications with their lawyers. The privilege can be claimed by the client, an authorized person, or the lawyer, unless no holder of the privilege exists Cal Evid Code § 954. Cal Evid Code § 912 addresses waiver of the privilege, stating it is waived if the holder voluntarily discloses a significant part of the communication, with exceptions for joint holders and privileged disclosures. Cal Evid Code § 917 presumes communications made in confidence during the attorney-client relationship are privileged unless proven otherwise.

In Nevada, the attorney-client privilege is primarily addressed in Nev. Rev. Stat. Ann. § 49.015, which establishes that no person has a privilege to refuse to disclose matters or prevent others from disclosing unless otherwise provided by law. In Texas, Tex. Evid. R. 503 outlines the attorney-client privilege, protecting confidential communications made to facilitate legal services, with exceptions for furtherance of crime or fraud Tex. Evid. R. 503. In New York, NY CLS CPLR § 4503 clarifies that beneficiaries of an estate are not considered clients of the attorney representing the personal representative, and fiduciary relationships do not waive the privilege NY CLS CPLR § 4503. NY CLS Rules Prof Conduct R 1.6 defines confidential information and restricts its disclosure without client consent, with exceptions for preventing harm or complying with legal obligations NY CLS Rules Prof Conduct R 1.6.

Should a privileged communication be disclosed to a third party, the privilege may be considered waived, resulting in the communication being susceptible to discovery during litigation or governmental inquiry. The attorney-client privilege is fundamentally designed to protect confidential communications between a client and their attorney, fostering open and honest discussions. It is rooted in public policy and aims to ensure effective legal representation. However, the privilege is not absolute and is subject to exceptions, such as when the communication involves furtherance of a crime or fraud. Additionally, the privilege does not extend to certain elements of the attorney-client relationship, such as the fact of representation, dates of representation, or fee arrangements

The inclusion of third parties in communications generally voids the privilege unless the third party qualifies as a representative of the client under specific legal definitions. For instance, in Nevada, representatives are limited to those authorized to obtain legal services on behalf of the client. This principle underscores the importance of maintaining confidentiality within the bounds of the attorney-client relationship.

II.The Intersection of AI Platforms and Privilege

Clients increasingly employ AI platforms to facilitate document drafting, seek legal information, review contracts, and even interface with counsel. AI-powered chatbots and document review solutions deliver efficiency and insight, yet they also present notable risks to confidentiality and privilege.

Categories of AI Platforms and Associated Risks

• Generative AI (e.g., ChatGPT, Google Bard): Clients may submit privileged information to generative AI platforms seeking legal advice or clarification regarding legal matters.
• Cloud-Based Legal Assistants: These services store, process, or manage client documents utilizing AI algorithms.
• Third-Party E-Discovery Platforms: Platforms leveraging AI to review evidence or communications.

A common concern is that, when clients transmit confidential legal information through these platforms, such data may be accessible to individuals outside the attorney-client relationship. This can precipitate inadvertent disclosure and, consequently, waiver of privilege.

III.Do AI Platforms Constitute Third Parties?

A critical legal issue is whether engaging an AI platform constitutes disclosure to a third party. Attorney-client privilege generally applies only to confidential communications between the lawyer and client, or their necessary agents. If a client provides privileged details to a platform managed by an entity not directly engaged by the law firm for legal support, courts may determine that privilege has been forfeited.

Relevant considerations include:

• Terms of Service and Privacy Policies: Numerous AI platforms’ terms permit the provider to review, utilize, or share user inputs with affiliates, contractors, or for service improvement.
• Absence of Confidentiality Obligations: Unlike attorneys, most technology providers are not bound by professional obligations to protect privileged information.
• Jurisdictional Variances: Statutes concerning privilege and waiver differ across jurisdictions, complicating cross-border legal matters.

Illustrative Scenarios

• A client uploads a contract draft containing legal advice to a cloud-based AI editing tool; if the tool retains or analyzes content externally, this may constitute disclosure.
• A business proprietor utilizes a generative AI chatbot to summarize a confidential legal memorandum; if platform policies sanction the use of input for model training, privilege may be compromised.
• An in-house counsel employs an AI-driven e-discovery platform vetted and retained by the law firm under robust confidentiality agreements; in this instance, privilege may be preserved.

VI.Legal Precedent and Emerging Standards

The current body of precedent regarding the intersection of AI technology and privilege waivers is still developing and nascent at best. However, courts have consistently held that voluntary disclosure of confidential communications to third parties not essential to legal representation results in forfeiture of privilege. Some courts have recognized that storing privileged communications with a cloud service provider does not result in waivers if reasonable measures are taken to preserve confidentiality, such as encryption and comprehensive contractual protections.

Conversely, use of public or consumer-facing AI services—characterized by ambiguous or unfavorable privacy policies—poses heightened risks. AI platforms frequently analyze, aggregate, and retain user data for purposes beyond mere storage, thereby increasing the potential for waiver.

The concept of implied waiver of attorney-client privilege applies to the use of AI platforms in legal contexts when confidential communications are disclosed to third parties, including AI tools, in a manner that undermines the confidentiality essential to the privilege. Courts have consistently held that the attorney-client privilege is waived, either expressly or by implication, when privileged communications are disclosed to individuals or entities outside the scope of the privilege, such as AI platforms that are not designed to maintain confidentiality United States v. Under Seal (In re Grand Jury Subpoena), 341 F.3d 331, In re Lott, 424 F.3d 446.

Implied waiver occurs when a party’s conduct, such as using an AI platform, places privileged information at issue or makes it relevant to a claim or defense. For example, if an attorney uses an AI tool to process or analyze privileged communications and the tool’s design or operation involves sharing the data with third parties, this could be construed as a voluntary disclosure, thereby waiving the privilege. Courts have emphasized that fairness principles guide the application of implied waiver, particularly when the privilege is used as both a shield and a sword, or when the opposing party is denied access to information vital to their case Dion v. Nationwide Mut. Ins. Co., 185 F.R.D. 288, Patrick v. City of Chicago, 154 F. Supp. 3d 705, UUSI, LLC v. United States, 121 Fed. Cl. 218.

Additionally, the use of generative AI tools, which rely on external data sources to generate content, poses unique risks to attorney-client privilege. These tools may inadvertently expose confidential communications to third parties or fail to safeguard the confidentiality of the information, leading to a waiver of privilege. Attorneys must exercise caution and ensure that any AI platform used complies with confidentiality requirements to avoid breaching the privilege.

V.Best practices for Legal Professionals and Clients

Considering these risks, both attorneys and clients must exercise vigilance and adhere to best practices to safeguard privilege:

i.Diligent Selection of Technology Providers

It’s our job as counsel to rigorously assess technology providers, including AI solutions, ensuring the following:

• The existence of comprehensive confidentiality agreements
• Secure, encrypted data storage and processing protocols
• Commitments from providers not to access, utilize, or disclose client data for unrelated purposes

While the American Bar Association’s Formal Opinion 512 issued on July 29, 2024, provides guidance to lawyers, there remains a dearth of guidance to clients, with respect to the ramifications of uploading confidential communications or attorney-work product to AI or LLM platforms.

ii.Clear Communication with Clients

Legal professionals should educate clients about the risks associated with using consumer AI platforms to store, analyze, or transmit privileged information. Establish explicit guidelines identifying approved tools and those to be avoided.

Where use of AI tools is indispensable, disclose only the minimum necessary information and avoid submitting unredacted privileged communications to public or unvetted platforms.

Both attorneys and clients must diligently evaluate the implications of these technologies, recognizing that convenience should not supersede the imperative of protecting attorney-client privilege. Through careful selection of technology providers, client education, and robust internal governance, legal professionals can embrace innovation while upholding one of the profession’s most fundamental protections.

As courts and regulatory authorities continue to address the ramifications of AI within the legal arena, perpetual vigilance and adaptability will remain essential. The privilege resides with the client, yet its preservation in an era dominated by digital technology is a shared responsibility—one that requires both technological acumen and unwavering dedication to confidentiality.