WHAT IS ONE PHRASE you would not want your attorney to say in court? Did the phrase “I’m here live, I’m not a cat”1 come to mind? Probably not, but attorney Rod Ponton nevertheless uttered this phrase to Judge Roy Ferguson as he struggled to remove a cat filter from his Zoom screen during a remote hearing held at the beginning of the pandemic.2 Luckily, Judge Ferguson kept his composure and chose not to punish an honest mistake made by a “senior lawyer doing his best in a very difficult time.”3 Most viewers watched the viral “lawyer-cat video” and simply chuckled at Ponton’s misfortune.4 However, lawyers are aware of the underlying warnings from the mistakes of their peers and should use “prospective hindsight” to avoid falling victim to similar fates.5 To lawyers, the comical lawyer-cat video is an allegory of the new risks technology poses to professional ethics. Indeed, a less forgiving judge might have interpreted the same scenario as a violation of professional ethics. The comical lawyer-cat video should lead all lawyers to reflect on their technological proficiencies and shortcomings.

Much of modern legal technology is far more complex than the video communication software attorney Ponton struggled with. In fact, many tasks traditionally “viewed as requiring human intelligence” are now often performed by artificial intelligence (AI) software.6. AI can perform legal research, data analysis, document drafting, and other functions conventionally executed by lawyers7. Proponents of AI acknowledge its capabilities in simplifying the lawyer’s tasks, such as due diligence.8 Nevertheless, critics are skeptical of a new technology that will effectively redefine the practice of law as we know it, particularly as it relates to ethics.9 But rather than resist the inevitable, lawyers should integrate AI into their practice and develop the requisite knowledge and skills to reap its benefits without jeopardizing their professional responsibilities.

As recognized in the Maryland Rules of Professional Conduct (the “Rules”), attorneys are agents of justice who must practice law competently and diligently and must maintain client confidences.10 Thus, when lawyers use AI without the requisite knowledge and skill to operate it competently and diligently, they essentially gamble with violating the Rules and risk compromising client confidentiality.11 If the current trend continues, AI will become standard practice, and even Luddites will be forced to succumb to technological advancements as a matter of professional responsibility.12 Humor aside, the time has come for all practitioners to reflect on the steady advancement in technology and its effects on the ethical practice of law.13

Rule 19-301.1 imposes an ongoing duty of competence and requires lawyers to “keep abreast of changes in the law and its practice.”14 Accordingly, lawyers who rely on AI should be cognizant of its benefits and limitations alike.15 One AI function that carries minimal risk regarding the duty of competency is automated document review.16 An attorney teaches the application to flag documents containing specified phrases or metadata.17 The process reduces time spent on repetitive tasks without posing a significant threat to legal professionalism, given that attorneys continue to use their practical legal knowledge to input relevant information for the AI software to identify, sustaining a degree of participation in the specific case.18

Conversely, the ethical stakes are higher when the world’s first AI lawyer, ROSS, drafts legal memoranda from start to finish.19 Lawyers who use AI to automate legal research and document drafting should understand the system’s capabilities and algorithms.20 AI research systems yield results by searching collections of legal authority assembled from source materials available in the public domain.21 These databases are manually built to promote efficiency by creating manipulated results based on practitioner needs.22 Efficient as it may be, limiting the scope of legal research to a tailored database can be a double-edged sword. The quality of AI-produced research is contingent upon the relevance, applicability, and scope of the legal authority available in manually constructed databases.23 An understanding of this foundational knowledge can help attorneys identify the need for additional research or errors in the system and ensure competency.24

Moreover, AI algorithms analyze repetitive human action and offer results based on the system’s perception of the user’s past needs.25 As such, algorithms can produce results that reflect user bias and poor search strategies.26 As attorneys serve clients with diverse backgrounds, it may be inappropriate to use AI when it could potentially bring bias into an attorney’s legal analysis.27 When an AI program uses data derived from human decisions containing unfair biases, the AI program could produce an outcome reflecting those unfair biases.28 The Department of Justice utilized an AI program meant to assess recidivism.29 However, the tool “overpredicted” the risk that people of color would commit crimes or violate rules after release compared to white people.30 AI algorithm experts suggest lawyers seek an “ethical algorithm design.31” However, an ethical algorithm design, which allows users to form algorithms that avoid integrating unwanted biases, can only combat the biases that users can identify.32 Ultimately, while AI is a valuable tool, attorneys must use it appropriately by maintaining a basic understanding of what occurs in its “black box” to appreciate all potential risks fully.33 However, an ethical algorithm design, which allows users to form algorithms that avoid integrating unwanted biases, can only combat the biases that users can identify.

In addition to representing clients competently, all attorneys must ensure that their communications with clients and their information remain confidential.34 Most relevant in the context of technological advancement is the obligation of maintaining client confidences, including client data35. Attorneys will share client data and communications when using a third-party AI program, potentially compromising that information.36 Data breaches involving technology stored or controlled by a separate entity pose unique risks to a lawyer’s obligation to safeguard client confidences.37 The duty to protect confidential information cannot be delegated to a third party, nor can it be avoided by virtue of an agency relationship between a lawyer and a third party.38 Lawyers using AI through a separate entity must ensure that their confidential information shared within the AI technology is not misused or is at a greater risk of breach than maintaining confidential information themselves.39

Fortunately, lawyers can pursue numerous options to reasonably protect confidential information shared with an AI provider.40 At a minimum, they should be aware of the provider’s confidentiality policies and procedures. Attorneys who use template software provided by third-party providers for document drafting may also consider the use of data encryption and authentication features as means of additional protection.41 While it may not be impossible to prevent confidentiality breaches in every instance, attorneys must be aware of the methods and means to mitigate any additional confidentiality risks associated with utilizing AI technology from a third party.42

Ready or not: the AI invasion is here, and competent use of it will soon be an integral aspect of practicing law.43 Instead of using unfamiliarity with AI to justify their mistakes, lawyers ought to accept that AI is here to stay and take steps to mitigate ethical risks.44 As technology evolves, lawyers must continually reimagine what it means to practice law competently and diligently.