Ethics Opinions

ETHICS OPINION NO. 267
OF THE MISSISSIPPI BAR
RENDERED November 14, 2024

CAVEAT: This opinion is limited strictly to the facts set forth in the hypothetical submitted and is limited to the question of whether the proposed conduct is permissible under the Mississippi Rules of Professional Conduct. The Ethics Committee is prohibited from rendering opinions on questions of law by Article 8-15(c) of the Bylaws of The Mississippi Bar. Any incidental reference to legal authorities is informational only and should not be taken as the Committee’s interpretation of such authorities or of the legal issues arising from the hypothetical presented or of the legal ramifications of the proposed conduct. The Committee’s opinion is limited to ethical issues only.

The Ethics Committee of the Mississippi Bar has been asked to render an opinion on the following:

Lawyer would like to use AI platforms and services in Lawyer’s practice and asks for guidance regarding whether Lawyer may ethically use such platforms, specifically:

  1. Does a lawyer have an affirmative duty to take reasonable measures and precautions to protect the confidentiality of client information when using generative AI?
  2. Does a lawyer have an affirmative duty to verify the accuracy and sufficiency of all work performed by generative AI?
  3. Does a lawyer have an affirmative duty to review costs and fees to ensure that any billing practices do not duplicate charges or inflate billable hours?
  4. And finally, does a lawyer have an ethical duty to disclose to the client that AI is being used with respect to legal matters entrusted to the lawyer by the client.

 

I.    Introduction 

Lawyers around the world including in Mississippi now make regular use of technology that incorporates Artificial Intelligence (AI) based technologies. AI includes computer technology, software, and systems that perform tasks traditionally requiring human intelligence. More recently, lawyers have started to use a new form of AI – Generative AI (GAI) – in their practice.  GAI can create new content, including text, images, audio, video, and software code in response to a user’s prompts and questions. GAI tools that produce new text are prediction tools that generate a statistically probable output when prompted from the dataset on which they are trained. Datasets are composed of large amounts of digital text culled from a variety of sources, including proprietary data sources in some cases. Some GAI tools are “self-learning,” meaning that as more data is included in the data set, the GAI will “learn” and evolve. GAI tools may assist lawyers in tasks such as legal research, contract review, document review, regulatory compliance, and drafting letters, contracts, briefs and other legal documents. 

AI and GAI tools offer the potential to help lawyers deliver higher quality and more efficient legal services by streamlining processes, analyzing large quantities of information, and drafting legal documents. In particular, AI can streamline legal research, review, analyze, and summarize large volumes of documents, automate repetitive tasks reducing manual labor, detect deception in emails or documents, predict case outcomes and legal trends based upon historical data, expedite responses to client inquiries, and reduce legal expenses to the client due to accelerated research and document preparation.  

There are also risks inherent in the use of this new technology. GAI may struggle to grasp complex legal concepts, producing inaccuracies and misinterpretations. GAI may “hallucinate” information when it perceives a pattern that is not actually present, including citations to nonexistent legal “authorities.” GAI models trained on biased data may perpetuate biases. GAI algorithms lack transparency making it difficult to understand how the software arrived at its conclusions. And of particular concern to the legal profession, the use of “self-learning” GAI tools risks the disclosure of client confidential information.   

This Opinion addresses ethical issues raised by the use of GAI in the practice of law but only those that have been presented to the Committee by formal request. The Committee’s analysis is necessarily general because there are a wide range of GAI tools that lawyers may choose to utilize.  Further, this new technology is changing rapidly, and its precise features and capabilities are evolving and will continue to change in ways that are difficult to anticipate.  

 

II.    A lawyer has an affirmative duty to take reasonable measures and precautions to protect the confidentiality of client information when using generative AI. 

Mississippi Rule of Professional Conduct 1.6(a) provides that “A lawyer shall not reveal information relating to the representation of a client unless the client gives informed consent….”  Confidential information is broadly defined to include all information relating to the representation.  The Comment to Rule 1.6 requires a lawyer to “act competently to safeguard information relating to the representation of a client against inadvertent or unauthorized disclosure by the lawyer or other persons who are participating in the representation of the client or who are subject to the lawyer’s supervision.”

When lawyers use GAI tools, they must be aware that those tools can place their client’s information at risk.  Trade secrets are of particular concern, but Rule 1.6 requires lawyers to protect all confidential information.  When lawyers input client information into GAI tools, they need to be aware of the risk that their clients’ confidential information could be disclosed to those who should not have access to it.  For example, some GAI tools use information that users input to train themselves, and those tools could disclose that information to other users in response to their prompts.

Lawyers can mitigate or even eliminate these concerns – and thereby fulfill their “duty to act competently to safeguard” confidential information – in a variety of ways.  First, lawyers should review the GAI tool’s terms of use (including privacy and data retention policies) to determine whether and how any information is shared with other users.  Some GAI tools do not pose a risk to confidentiality.  Similarly, lawyers may be able to negotiate terms with providers that ensure that client confidential information is protected.

Second, if possible, lawyers should not input identifiable client information into the GAI tool.  See Mississippi Rule of Professional Conduct 1.6 Comment (“A lawyer’s use of a hypothetical to discuss issues relating to the representation is permissible so long as there is no reasonable likelihood that the listener will be able to ascertain the identity of the client or the situation involved.”).  Of course, depending on the tool, this might not be possible as the effectiveness of some GAI tools depends on receiving client information.

If a lawyer intends to use a GAI tool and remains concerned that, despite taking reasonable precautions to prevent the disclosure of confidential information, that information is at risk, the lawyer should obtain the client’s informed consent.  To obtain the client’s informed consent, the lawyer must “discuss with the client the proposed use of AI, the applications of AI to be utilized, the risks and benefits of the AI product, and fully explain privacy concerns.”   

Consistent with Mississippi Rules of Professional Conduct 5.1 and 5.3, lawyers with managerial and supervisory responsibilities must also make efforts to ensure that subordinate lawyers and non-lawyers (whether employed by the firm or third-party contractors) not reveal and take reasonable measures to protect confidential client information.  Accordingly, firms should train their legal and non-legal staff on these issues and consider firm policies addressing these issues.

 

III.    A lawyer has an affirmative duty to verify the accuracy and sufficiency of  work performed by generative AI. 

Mississippi Rule of Professional Conduct 1.1 requires that a lawyer “provide competent representation to a client.  Competent representation requires the legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation.”  The Comment to Rule 1.1 requires that lawyers “maintain the requisite knowledge and skill” through “continuing study and education.”  The Committee has previously recognized that keeping up with technology and understanding the benefits and risks of new technology are an important part of the lawyer’s duty of competence.  See Ethics Opinion No. 263 (“lawyers must weigh the benefits of cloud-based storage against the new risks presented by that technology.”); Ethics Opinion No. 259 (recognizing that lawyers have an ethical duty to understand how metadata is created).  AI technology is changing rapidly, and lawyers need to keep up with those changes in order to competently represent their clients.  

Consistent with Mississippi Rules of Professional Conduct 5.1 and 5.3, lawyers with managerial and supervisory responsibilities must also make efforts to ensure that subordinate lawyers and non-lawyers (whether employed by the firm or third-party contractors) are competent with GAI.  

As with other aspects of law practice, there are several different paths to achieving competence.  Lawyers may achieve the required competence by “engaging in self-study, associating with another competent lawyer, or consulting with an individual who has sufficient expertise in the relevant field….This means that lawyers should either acquire a reasonable understanding of the benefits and risks of GAI tools that they employ in their practices or draw on the expertise of others who can provide guidance about the relevant GAI tool’s capabilities and limitations.”   

GAI tools offer significant potential benefits to clients in the form of higher quality and more efficient legal services.  At the same time, however, the use of these tools creates risks.  Among the most concerning is that GAI tools sometimes create inaccurate output, a concern noted by Chief Justice John Roberts in his 2023 Annual Report on the Judiciary.   In several widely reported cases, GAI tools have “hallucinated” legal authorities, and lawyers who carelessly relied on the fictional authorities have been sanctioned.

Obviously, then, lawyers cannot indiscriminately or solely rely on GAI tools, but if lawyers must use other methods to repeat all the work performed by the GAI tools, then there is no point in using those tools in the first place.  When a lawyer uses a GAI tool to perform a task, he should consider that tool as a “virtual assistant” and treat the work product of that tool similar to how he would treat work product performed by subordinate lawyers or nonlawyers: the lawyer remains fully responsible for the work, and the proper approach is “trust but verify.”    As part of the lawyer’s responsible use of a GAI tool, the lawyer should determine whether there is a reasonable basis for trusting the tool’s output. GAI tools perform a wide variety of tasks, and the level of trust and amount of verification depends on the circumstances.  We adopt the ABA Ethics Committee’s analysis of this issue:

The appropriate amount of independent verification or review required to satisfy Rule 1.1 will necessarily depend on the GAI tool and the specific task that it performs as part of the lawyer’s representation of a client. For example, if a lawyer relies on a GAI tool to review and summarize numerous, lengthy contracts, the lawyer would not necessarily have to manually review the entire set of documents to verify the results if the lawyer had previously tested the accuracy of the tool on a smaller subset of documents by manually reviewing those documents, comparing then to the summaries produced by the tool, and finding the summaries accurate. Moreover, a lawyer’s use of a GAI tool designed specifically for the practice of law or to perform a discrete legal task, such as generating ideas, may require less independent verification or review, particularly where a lawyer’s prior experience with the GAI tool provides a reasonable basis for relying on its results. 

Moreover, lawyers may not abdicate their responsibilities by relying solely on GAI tools to perform tasks that call for the exercise of professional judgment” such as “offer[ing] legal advice to clients, negotiat[ing] clients’ claims or perform[ing] other functions that require a lawyer’s personal judgment or participation. 

 

IV.    A lawyer has an affirmative duty to review costs and fees to ensure that any billing practices do not duplicate charges or inflate billable hours.  

Mississippi Rule of Professional Conduct 1.5(a) requires that all of a lawyer’s fees and expenses be “reasonable” and includes a non-exclusive list of criteria for evaluating whether a fee or expense meets that standard. Rule 1.5(b) requires a lawyer to communicate to a client the basis on which the lawyer will charge for fees and expenses (unless the client is a regularly represented client and the terms are not changing). The required information must be communicated before or within a reasonable time of commencing the representation, preferably in writing. Therefore, if a lawyer wants to treat the GAI tool as an expense and charge the client for the use of the GAI tools or services, the lawyer must explain the basis for the charge within a reasonable time of commencing the representation, preferably in writing. 

This logic applies if the lawyer uses the GAI tool himself or uses a third-party provider’s GAI service.  Thus, for example, if the lawyer uses a third-party provider’s GAI tool to review thousands of voluminous contracts for a particular client and the provider charges the lawyer for using the tool on a per-use basis, it would ordinarily be reasonable for the lawyer to bill the client as an expense for the actual out-of-pocket expense incurred for using that tool, provided the lawyer communicates this to the client.

If, on the other hand, the lawyer chooses to treat GAI as an overhead expense, that is a different matter.   Lawyers may not bill clients for general office overhead expenses, including the routine costs of maintaining a library, securing malpractice insurance, renting office space, and the like. Such overhead should be subsumed within the lawyer's charges for professional services.  Accordingly, absent disclosure to the client, a lawyer should consider the cost of GAI tools and services to be overhead that may not be charged to the client. 

Lawyers must also consider the impact that the use of GAI tools may have on the time they expend to complete certain tasks. One of the chief benefits of GAI tools is that they may allow lawyers to deliver legal services in a faster and more efficient manner.  The benefit of any efficiency achieved must go to the client, not the lawyer.  For example, lawyers who bill clients on an hourly basis for time spent on a matter may only bill for their actual time, and not more.  Thus, regardless of the efficiencies achieved with GAI tools, the client should only be charged a reasonable fee for the legal services performed which means that the lawyer is only compensated for time actually and reasonably expended. 

Further, lawyers should not bill their client for time spent to gain knowledge about GAI tools and services. Costs incurred in learning about GAI, in maintaining GAI-provided services, and keeping up to date with changes in its use, should be considered as a part of the lawyer’s overhead, just like any other continuing legal education expense. 

Moreover, the lawyer who charges a flat or contingency fee must also ensure the reasonableness of those fees when the lawyer uses a GAI tool.  Because the use of a GAI tool may enable the lawyer to complete tasks more quickly than without the tool, it may be unreasonable for the lawyer to charge the same flat or contingency fee when using the GAI tool as when not using it. 

The best practice is to memorialize any agreement that the lawyer and the client have regarding billing practices or billing terms relating to GAI tools and services in writing before billing the client.

 

V.    A lawyer has an ethical duty to disclose to the client that Al is being used with respect to legal matters entrusted to the attorney by the client in certain circumstances. 

Mississippi Rule of Professional Conduct 1.4 requires that a lawyer keep the client reasonably informed about the status of their matter, to promptly inform the client of any decision or circumstance which requires the client’s informed consent, and to obtain the client’s informed consent of such decision or circumstance. Further, the attorney is required to “reasonably consult” with the client about the means by which the client’s objectives are to be accomplished. 

Routine use of a GAI tool in a client’s matter does not in and of itself require specific communication to the client, unless the client is being charged for the costs of the GAI tool, there is a risk of disclosing confidential client information, or if the disclosure of the use of the GAI tool is required by court or other rules. However, certain situations will require disclosure in order to comply with Rules 1.5 and 1.6.  When a lawyer will be providing confidential client information to a third-party provider, when the lawyer will be passing the expense of the AI tool on to the client, or when the lawyer is concerned that despite taking appropriate preventative measures confidential information may be inadvertently disclosed, then the lawyer must obtain the client's informed consent.


Conclusion

Lawyers in Mississippi may ethically use Generative AI, provided the lawyer has made appropriate safeguards to protect client confidential information, is competent to use the technology, takes precautions to verify the accuracy of the tool’s output, uses reasonable billing practices, and obtains the client’s informed consent when appropriate.