Trust, But Verify: Use of AI and the Estate Lawyer’s Duty

By George Riter, Esq., Chair, Probate & Tax Section, and Elizabeth Ferraro, Esq., Secretary, Probate & Tax Section

Catchy title, right? ChatGPT generated 12 options for quippy titles after a two-sentence explanation of this topic and Sidebar’s target audience. Most readers likely consider this a harmless use of Artificial Intelligence (AI) that can save our already over-taxed brains a little thinking power. However, while AI is everywhere, it is certainly not all harmless.

Outlook can boil down your client’s lengthy email to a few points and comment on your grammar; Google will write the response email for you; and chatbots can follow a prompt to provide a response, known as generative AI. While each of these items can save time, each comes with significant risks to attorneys’ ethical duties.

This article touches on the professional responsibilities owed to clients as our trusts and estates practices intersect with the use of AI and proposes that proper training, along with disclosure in engagement letters, is the best course to protect both clients and attorneys.

Pennsylvania Rules of Professional Conduct

When using AI in practice, the duties of competence and confidentiality are some of the most vulnerable. The Pennsylvania Rules of Professional Conduct, the “Rules,” provide the following relevant guidelines in these areas.

• Maintaining Competence: Rule 1.1(8) provides that a lawyer “must keep abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology . . . .”
• Confidentiality: Rules 1.6(a) and 1.6(d) provide that a lawyer “shall not reveal information relating to representation of a client unless the client gives informed consent, except for disclosures that are impliedly authorized in order to carry out the representation . . . .” In addition, a “lawyer shall make reasonable efforts to prevent the inadvertent or unauthorized disclosure of, or unauthorized access to, information relating to the representation of a client.” This “duty not to reveal information relating to representation of a client continues after the client-lawyer relationship has terminated.” Rule 1.6(e).

Lawyers in general, and estate lawyers in particular, tend to be set in their ways. However, Rule 1.1(8) mandates that lawyers understand the benefits and risks of their use of technology, including AI.

Generative AI that produces its own content based on prompts or input provided by the user is a tempting time saver, but risks providing unreliable results, as seen often in litigation. For example, you could ask an AI platform to draft a will compliant with Pennsylvania law that creates a trust for minor children. It is your professional responsibility to weigh the upside of using this technology, look how much time and money you saved your client, against the downside. AI is not a lawyer. Query whether the document is legally sound and if it meets client expectations.

When examined by various courts, most jurists will examine the reasonableness of the attorney’s use of AI when determining whether such use was proper.

Now consider whether inputting client information into AI is a violation of Rule 1.6 relating to confidentiality. AI tools are not clients, and accordingly, providing attorney-client-privileged information to these systems may result in a loss of the attorney-client privilege or privileges over any work product provided. Similarly, clients entering attorney communications may also be deemed to have waived privilege.

Generative self-learning AI services, such as ChatGPT, learn from users’ input, meaning they derive “knowledge” from the information gathered from users around the world. As a result, if using external generative AI, i.e., not a closed system, a line from your email could appear in the answer to another user’s prompt.

Although Pennsylvania has no precedent, the Southern District of New York determined that a client’s prompts to a generative AI system, along with documents generated by AI to share with counsel, were not protected by the attorney-client privilege or the work product doctrine. Joint Formal Ethics Opinion 2024-200 also provides non-binding guidance for practitioners using AI, although, since its publication in 2024, the technology has significantly advanced.

Ultimately, information provided to AI services may not remain solely between you and the AI platform and may instead become part of the greater AI domain. This depends on the platform’s terms and conditions. In addition, many businesses are developing their own closed-model AI systems or encouraging employees to use AI that comes with secure software programs, like Microsoft’s Copilot. Competence and awareness of the security of the service are imperative for maintaining confidentiality and privilege.

Practice Points

We all strive to become more efficient and better at what we do. AI, like any other technology, can be a useful tool. As attorneys, we need to consider reasonable ways to incorporate AI into our practice, disclose the use to clients as necessary, and be aware of the risks posed to both clients and our professional responsibilities.

When incorporating AI into our practice, we should all consider these practice points:

• Determine what areas of your workload can be made more efficient using AI without disclosing confidential information or posing risks to the client. Develop and regularly review a firm policy that outlines the acceptable uses for AI. Ensure everyone acknowledges and follows the policy.
• Do your due diligence in procuring AI products to determine what safeguards are in place to secure your client’s confidential information. For example, if using a digital notetaker, it is your duty to determine how the service uses and stores the information it gathers.
• In your engagement letters, adequately disclose the use of AI to your clients and provide clients with the opportunity to opt out. Also inform clients of the risks to confidentiality and privilege posed by using generative AI outside of a closed-model system to review correspondence or documents you have prepared or to generate a response.
• The model rules of professional conduct provide that competence in technology cannot be merely satisfied by hiring a qualified IT provider. Attorneys must educate themselves on both the risks and benefits of the technology. This can be accomplished through self-education or CLE programming.

Frequently Asked Questions

Q: Are there specific AI platforms or tools you currently recommend, or caution against, for trusts and estates practitioners in Pennsylvania, and why?

A: As this area develops, we are advising that practitioners stick to closed systems that pull only from legitimate sources.  For example, legal research tools that have created AI platforms, such as Lexis and RIA Checkpoint, which only use verified sources and provide citations for their results.  Our Firm does not currently use AI to draft legal documents.  Furthermore, anyone using AI must verify the accuracy of the material.

Q: What should a compliant AI disclosure clause actually look like in an engagement letter, can you provide sample language?

A:

Sample

Regarding Attorney Use of AI During the course of our representation the Firm may use generative artificial intelligence (“AI”) for legal research.  The Firm will not engage in the use of generative AI for purposes of drafting or reviewing legal documents without additional disclosure to you.

Sample

Regarding Client Use of AI

NOTE

that the use of generative artificial intelligence (“AI”), including inputting our communications to you, our work product, or your communications to the firm, in such software or web-based browsers, can cause our communications to lose their privileged (i.e., protected) status.  The firm does not authorize the disclosure of our communications or work product into any AI software and you should discuss with me prior to utilizing any such technology in connection with the firm’s representation of you.

Q: If a client has already used a public generative AI tool to draft their own will or trust document before retaining counsel, what are the attorney’s obligations upon reviewing it?

A: As with any client that comes to us with existing estate planning documents, we review the documents and advise if updates or other changes are recommended to accomplish the client’s goals.  As a general rule of thumb, we do not advise relying on AI to draft estate planning documents.

Q: How should Pennsylvania estate attorneys handle AI-generated documents in the context of court filings or submissions to the Register of Wills, are there any jurisdiction-specific requirements emerging?

A: The Register of Wills and Orphans’ Court have specific forms that can be used at least as a template for a number of filings.  There are specific rules in practice with the Register of Wills and Orphans’ Court that we are familiar with as attorneys that AI may not be compliant with.

Q: Does using AI to take notes or summarize communications during client meetings, without explicit consent, create a separate set of ethical concerns beyond those raised by generative AI drafting tools?

A: Any platform for virtual note taking should be properly vetted to make sure that client’s information is secure and that the use of any such service does not result in any such conversations losing their privileged status.