With the increased popularity of AI, it is bound to make waves in just about any profession. The legal profession is no different. The overwhelming consensus among legal professionals is that AI is a great help rather than a hinderance. This article’s purpose is to outline the proper uses for AI in the legal profession and detail what kind of problems legal professionals will need to confront if they introduce AI into their practice.
AI or artificial intelligence is a system much like Google in that you can search it for almost anything and it will be able to provide you with an answer. There are two different kinds of AI—generative and agentic. Generative AI is ChatGPT or Google AI, which are engines that you can prompt to create things or answer questions based on things that already exist. Generative AI is not necessarily known for or used for original ideas or complex thinking. Agentic AI is a more advanced system which can handle complex tasks and even act on its own in reference to what you request. Agentic AI is the system that legal professionals should look to if they are going to employ AI in their practice.
Westlaw polls legal professionals every year on a myriad of topics. When polling them about AI, 77% of respondents stated that they use AI for document review, 74% use it to summarize documents and research, and 59% use AI to aid them in drafting memos and briefs. Legal professionals also reported that they anticipate a decline in billing models over the next 5 years because of how much time AI saves. Client relationships stand to reap great benefit from their lawyer using AI because lawyers charge in increments of time. So, if an attorney is spending 2 hours on your case with AI in lieu of taking 5-6 hours independently, they are saving their client quite a bit of money.[1]
The Model Rules of Professional Conduct (MRPC) are implicated when discussing the use of AI because lawyers have ethical standards they must adhere to. Worries about violating the MRPC by incorporating AI can be laid to rest though, because the American Bar Association (ABA) has issued opinions in support of its use in the legal profession. The ABA stated that instead of trying to change the MRPC to suit the use of AI, we can instead use the rules themselves to govern its use. They explained that the rules were written to have flexibility that will make them easier to apply as society and technology evolve.
Lawyers must exercise competence, candor, and communication. Additionally, they must maintain confidentiality with their clients. Michael Murray, a professor at the University of Kentucky expands on this thought in his article, Algorithmic Ethics in an Era of Agentic AI Advocacy. There, he explains how AI bolsters the MRPC in some respects and highlights possible points of tension.[2]
Competence is a foundational rule on which proper and ethical lawyering rests. To be a competent lawyer, a lawyer must keep abreast with the changes around them. This includes, per the ABA, the use of AI. With the introduction of AI, competence has evolved to require lawyers to not only be aware of its existence, but they must establish a “reasonable and practical understanding” of its functions and limitations.
Confidentiality is of the more sensitive rules when using AI because it could result in an inadvertent disclosure of confidential information. This can be occur differently depending on what kind of AI system is being used. Generative AI, as described by Murray, impacts the task of legal practice, whereas Agentic AI affects personal judgment and autonomy. The risk of breach is stronger for Agentic AI, because it needs less human input than other AI models. Its semi-autonomous actions can end up violating confidentiality because it can self-correct and adjust course on its own based on its findings.
When selecting a system, a lawyer can choose between a Non-Privacy Protecting system or a Locked Down system. The former may use user inputs for training without prior notice or permission, while the latter is designed for legal practice, but prioritizes data security. To narrow their choices, Murray suggests that lawyers read and understand each AI system’s terms of service, privacy policy, and data security protocols. From there, a lawyer should select a privacy protecting system that can guarantee client data will be kept confidential. If client data is not guaranteed to remain confidential, the lawyer must then take steps to anonymize the data so there is no breach. In doing that, the client must give informed consent because the use of their information in the AI system could later cause a breach.[3]
One of the most basic rules a lawyer is required to adhere to is communication. Typically, lawyers are not required to share each minute task they perform for a client’s case. If they are using AI, that standard shifts because now another “being” is participating in their representation. If you are using AI to do tasks that your client assumes would be performed by you, disclosure of AI use is likely required. This principle follows and is applied similarly to potential passed on costs from AI use. A lawyer must disclose the use of AI and acquire informed consent from the client. The same principle follows billing a client for tasks performed by AI.
The ABA addressed this issue on point in their Formal Opinion 512, stating “Absent an agreement, the firm may charge the client no more than the direct cost associated with the tool (if any) plus any reasonable allocation of expenses directly associated with providing the AI tool.”[4] Once consent to the associated costs or fees is established, the lawyer can move through the case with AI and have no issues. A lawyer may not charge for the time and education it may take for them to become competent about AI.
Formal Opinion 512 goes on to address the duty of verification, which became a hot topic recently. This is because lawyers were not exercising competence when using AI to draft citations, and they failed to verify that the cases they cited existed. The attorneys for Al Hamim v. Star Hearthstone out of Colorado received stern warning of future punishment if they submitted another filing with hallucinated citations or cases. The opinion states “lawyers must critically review, independently verify, and take full responsibility for all AI-generated content before it is submitted to any tribunal.”[5]
All in all, AI has made tremendous strides in the legal profession and has garnered support from a large population of legal professionals, including their regulatory body, the American Bar Association. This support and employment of AI in the legal profession is a positive, but it will only remain positive if lawyers are able to maintain their ethical and professional responsibilities while using it. As long as a lawyer is vetting the AI system, prioritizing confidentiality and privacy protection, and communicating with their client about their usage, AI can be an asset that leads to more effective representation.[6]