Introduction

Generative AI (Gen AI)[i] tools are increasingly being integrated into legal practice.[ii] While they offer significant benefits for efficiency and productivity, their use comes with important ethical and professional considerations. This short guide outlines key best practices based on guidance from the American Bar Association[iii], state bar associations[iv] (including Tennessee) and court rulings.

Confidentiality, Competence & Client Consent

• Duty of Confidentiality: Lawyers must safeguard all client information. This duty applies fully to Gen AI use.[v]
  • In practice, this means not uploading or disclosing confidential data into an unsecure or non-private Gen AI tool. Have an internal protocol with a quick checklist to ask questions like: Will this prompt reveal client identity? Is there another way to phrase it? Am I uploading client documents?[vi]
• Duty of Competency: Lawyers should understand the benefits and risks associated with technology.[vii]
  • In practice, this means knowing what your Gen AI tool can and can’t do. Training on how to use a tool effectively, verifying the accuracy of outputs and understanding the value and limitations of usage are all essential to providing competent representation.  
• Informed Client Consent: Obtain specific permission before inputting any client information into Gen AI tools. This consent is also needed if using an AI tool will significantly influence the representation. Boilerplate consent in engagement letters may be insufficient.
  • In practice, this means updating your engagement letters to disclose which Gen AI tools are being used and why, in plain language. This can also include the risks and safeguards you have taken in choosing a particular product. Allow clients to opt out of you using their data in Gen AI tools. If the tool is being used to make important decisions in the representation, such as settlement valuation or jury selection, get consent.
• Data Minimization: Share only necessary information with AI tools, removing names, identifying details and confidential information whenever possible.
  • In practice, this means stripping any identifying information before use. Instead of uploading the entire complaint, work with smaller, de-identified sections of text. 
• Security Vetting: Use only AI tools with robust encryption, clear privacy policies and confidentiality guarantees. Avoid services that claim ownership of input/output data or lack adequate security measures.[viii]
  • In practice, this means your firm has a due diligence process in place prior to any acquisition of new Gen AI technology. Inquire about contractual guarantees for ownership of inputs, outputs, and notice or indemnity provisions in the event of a breach, etc. Consider having an “approved tool” list and block other applications with a firewall.
• Privilege Preservation: Take precautions to ensure that sharing information with Gen AI tools doesn't jeopardize attorney-client privilege.[ix]
  • In practice, this means using a workspace or program that only you have access to. Avoid shared or public chat histories. Retain outputs in the client files and apply the same policies that govern emails and drafts. Avoid mixing privileged and non-privileged prompts in the same chat. Recheck the terms of service whenever updated by your AI provider to ensure privilege considerations are not impacted.

Legal Research and Writing Applications

AI as Assistant, Not Expert

• Use AI as a research and drafting assistant, not a substitute for legal judgment. The attorney remains responsible for all work product, including checking its accuracy.
• Leverage AI for efficiency (summarizing documents, generating draft memos, identifying patterns, etc.) while applying your own legal expertise to evaluate results. Using AI is just one step in the overall process of producing work product. 

Verification Requirements

• Mandatory Accuracy Checks: Verify all AI-generated content before use. AI can "hallucinate" non-existent cases, statutes, or misrepresent holdings.  Not only is verifying the case or statute important, you must go further and ensure all quotes and stated reasons for including the authority are accurate.
• Court Expectations: A growing number of judges require disclosure when Gen AI is used for drafting or research and require attorneys to certify that the work has been reviewed and verified by a human. Always check your local rules and the judge’s individual practices with regard to AI usage and disclosure.[x]  

Research Limitations

• Many publicly available AI tools have fixed knowledge cutoffs and, unless paired with services like Lexis or Westlaw, lack comprehensive and current legal content. It is essential that you understand the data that your Gen AI program has access to and is using when producing outputs. These sources could include the internet, case law databases or internal firm documents.
• Use AI to complement, not replace, traditional legal research using up to date resources (Westlaw, Lexis, etc.). An advantage to AI for research is its ability to distill concepts and search semantically, not just relying on exact search terms. While AI can help locate authority, it is still your role to determine if it is in fact authoritative, accurate and appropriate for your given matter.

Combine AI's efficiency with human expertise for optimal results. It is important to remember that widely available generative AI tools have only been on the market for roughly three years. The legal profession is still in the early stages of exploring and adopting this ever-evolving technology.

[i] George Lawton, What is GenAI? Generative AI explained, TechTarget (Mar. 13, 2025), https://www.techtarget.com/searchenterpriseai/definition/generative-AI.

[ii] Thomson Reuters Institute, 2025 Generative AI in Professional Services Report, Thomson Reuters (Apr. 15, 2025), https://www.thomsonreuters.com/content/dam/ewp-m/documents/thomsonreuters/en/pdf/reports/2025-generative-ai-in-professional-services-report-tr5433489-rgb.pdf.

[iii] ABA Standing Comm. on Ethics & Prof’l Responsibility, Formal Opinion 512: Generative Artificial Intelligence Tools, Am. Bar Ass’n (July 29, 2024), https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf.

[iv] AI and Attorney Ethics Rules: 50-State Survey, Justia (Apr. 2025), https://www.justia.com/trials-litigation/ai-and-attorney-ethics-rules-50-state-survey/.

[v] Tenn. Sup. Ct. R. 8, RPC 1.6 (2023), https://www.tncourts.gov/courts/supreme-court/rules/supreme-court-rules/rule-8-rules-professional-conduct; Formal Op. 512, supra note iii.

[vi] Ilia Kolochenko, How to Protect Your Law Firm’s Data in the Era of GenAI, Bus. L. Today (Dec. 9, 2024), https://www.americanbar.org/groups/business_law/resources/business-law-today/2024-december/how-protect-law-firm-data-era-gen-ai/.

[vii] Tenn. Sup. Ct. R. 8, RPC 1.1 cmt. 8 (2023).

[viii] Oliver Roberts, Legal AI Unfiltered: Legal Tech Execs Speak on Privacy and Security, Nat’l L. Rev. (Mar. 31, 2025), https://natlawreview.com/article/legal-ai-unfiltered-legal-tech-execs-speak-privacy-and-security.

[ix] Christopher Campbell & Antonious Sadek, Protecting Attorney-Client Privilege and Work Product in a GenAI World, Practical Guidance J. (Mar. 20, 2024), https://www.lexisnexis.com/community/insights/legal/practical-guidance-journal/b/pa/posts/protecting-attorney-client-privilege-and-work-product-in-a-generative-ai-world.

[x] Ropes & Gray LLP, Artificial Intelligence Court Order Tracker (last visited July 2, 2025), https://www.ropesgray.com/en/sites/artificial-intelligence-court-order-tracker.