Use this button to switch between dark and light mode.

Copyright © 2024 LexisNexis and/or its Licensors.

Generative AI in Discovery: GPT Prompt Preservation and Production Best Practices

December 08, 2023 (5 min read)

By: Rose J. Hunter Jones, Kassi R. Burns,  and Meredith A. Perlman, KING & SPALDING

THIS ARTICLE DISCUSSES BEST PRACTICES AND STRATEGIC INSIGHTS LITIGATORS SHOULD CONSIDER IN A FEDERAL COURT LITIGATION when dealing with discovery produced by generative artificial intelligence (GAI) tools.

Since the launch of OpenAI’S ChatGPT 3.5 in November 2022, discussions about the use of GAI tools like ChatGPT have dominated the news, publications, conference panels, and social media. The use of generated pre-trained transformers (GPT) is a topic that spans all industries and verticals, with the practice of law being no exception.

While the legal profession grapples with heady topics like appropriate use cases, potential bias, privilege and confidentiality considerations, and the application of legal ethics when using GPT and other GAI tools, it is important to look ahead to contemplate how the use of these tools will be addressed in litigation. This article focuses specifically on preservation and production obligations in federal civil litigation and how those obligations may apply to the emerging GPT tools businesses and individuals increasingly are using.

Finally, this article also contemplates the development of corporate GAI policies that will address a wide variety of issues related to the use of GPT tools. For example, these policies could include requirements for employees to acknowledge the use of GPT tools as a step in their process, which will ultimately be validated, refined, and finalized by the employee to accurately reflect the intended message or information.

For practical guidance on GPT prompt engineering, preservation obligations, production obligations, and practical considerations, subscribers may follow this link. Non-subscribers please sign up for a free trial of Practical Guidance to read this complete article.

Preparing for Our GAI Future in Litigation

In the context of federal civil litigation, addressing preservation and production obligations regarding GPT and prompts is essential for maintaining the integrity of legal proceedings. The evolving landscape of GAI, particularly GPT tools, introduces unique challenges that legal professionals, businesses, and individuals must navigate.

Preservation obligations encompass the duty to protect and retain relevant ESI in anticipation of or during litigation. GPT-generated content and prompts, being a form of ESI, require special attention due to its characteristics, such as identification challenges and data integrity concerns.

Production obligations come into play when parties need to produce ESI as part of discovery requests. GPT-generated data and prompts may be subject to these obligations, necessitating careful consideration of relevance, proportionality, and production format.

In conclusion, the integration of GPT tools into litigation practices requires a nuanced understanding of preservation and production obligations. It calls for the development of best practices and the incorporation of expert consultation. Legal professionals, businesses, and individuals must adapt to the unique challenges presented by GPT-generated content and prompts while upholding the principles of fairness, transparency, and compliance with legal obligations in federal civil litigation. 

Subscribers may view the full article in Practical Guidance. 

Not yet a Practical Guidance subscriber? Sign up for a free trial of Practical Guidance to read the full article.


Rose J. Hunter Jones is a partner at King & Spalding and leads the firm’s nationally ranked E-Discovery practice. Rose is dedicated to understanding and managing the intersection of legal and technical issues that now largely dominate American and global litigation, data breach response, and government investigations. She devotes her entire practice to crisis management, eData, and technology. As a crisis manager with 20 years’ experience, Rose works closely with clients legal, information technology, compliance, and records management departments to strategize, manage and defend corporations when facing the most complex and high-value matters. In addition, she is actively engaged in e-discovery thought leadership and shaping the law around emerging issues.


Kassi R. Burns is a senior attorney in King & Spalding’s E-Discovery practice. She is a skilled e-discovery practitioner with more than 13 years of progressive experience advising corporate clients and legal professionals on e-discovery best practices and the integration of technology and the law. Her diverse background ranges from transactional law, e-discovery (technical and managed review), and legal operations.


Meredith A. Perlman is an attorney in King & Spalding’s E-Discovery practice. She has 15 years’ experience across multiple review platforms and primarily focuses her practice on complex discovery issues, including the electronic discovery process. Meredith is experienced in all stages of case review and preparation using analytics and active learning to achieve the most efficient and best results for clients.


Related Content

For a comprehensive guide to current practical guidance on generative artificial intelligence (GAI), ChatGPT, and similar tools, see

GENERATIVE ARTIFICIAL INTELLIGENCE (AI) RESOURCE KIT

For a discussion of ethical issues faced by attorneys when considering using GAI technology in their practices, see

AI AND LEGAL ETHICS: WHAT LAWYERS NEED TO KNOW

For insight into a judge’s view of the use of GAI, see

ARTIFICIAL INTELLIGENCE: A JUDGE’S VIEW OF GENERATIVE AI

For an overview of the primary issues relating to the use of ChatGPT in the practice of law, see

LAWYERS AND CHATGPT: BEST PRACTICES

For an introduction to predictive coding in the context of discovery in federal court litigation, see

PREDICTIVE CODING FUNDAMENTALS (FEDERAL)

For recommendations on best practices to be used by litigators when working with e-discovery in federal court litigation, see

E-DISCOVERY BEST PRACTICES (FEDERAL)

To explore the role of the metadata of electronically stored information in e-discovery in federal court litigation, see

METADATA IN E-DISCOVERY (FEDERAL)

For an analysis of technology-assisted review of electronically stored information, see

TECHNOLOGY-ASSISTED REVIEW: OVERVIEW (FEDERAL)

For a sample certificate regarding the use of GAI in federal court, see

GENERATIVE ARTIFICIAL INTELLIGENCE (AI) USE AND COMPLIANCE CERTIFICATION (FEDERAL)

For a comparison of the most common electronically stored information production formats found in federal court civil litigation and the advantages and limitations of each format, see

ESI PRODUCTION FORMATS CHART (FEDERAL)

For an explanation on how email threading works, the technological benefits of using threading, and strategic considerations for it in document review and production in federal court litigation, see

EMAIL THREADING IN E-DISCOVERY CHECKLIST (FEDERAL)

For a review of issues relating to e-discovery search terms and search parameters in federal court litigation, see

E-DISCOVERY SEARCH TERMS CHECKLIST (FEDERAL)