[co-author: Jennifer Williams]

Privilege logging is one of the most time-consuming, expensive, and contentious components of discovery. Federal Rule of Civil Procedure 26(b)(5)(A), which governs the withholding of privileged material, provides little concrete guidance to alleviate these burdens. But the broad language of the rule also offers litigants significant flexibility to incorporate new technology, most notably automation and artificial intelligence, into their privilege logging practices. These tools meaningfully increase efficiency and cost savings for many litigants and should be more widely adopted as the technology progresses.

The Federal Rules

When a party withholds allegedly privileged material, Rule 26(b)(5)(A) requires that party to “expressly make the claim” of privilege and “describe the nature” of withheld documents “in a manner that, without revealing information itself privileged or protected, will enable other parties to assess the claim.” Parties may satisfy these requirements in many ways, but the four types of privilege logs listed below are the most common.¹ To best comply with Rule 1’s mandate “to secure the just, speedy, and inexpensive determination of every action,” elements of each type of log should be combined and automated according to the circumstances of the specific case.

Using Automation in Privilege Logging Today

• Traditional privilege logs typically contain information about a document’s author, recipient, dates, subject matter, narrative descriptions of the document’s content, and other information sufficient for the other party to assess the claim of privilege. Because attorneys typically create these logs on a document-by-document basis, they can be particularly burdensome and expensive. By definition these logs are not automated. Today they are rarely used in large litigation matters, but courts still sometimes offer litigants this option, particularly for hard copy documents.²
• Metadata privilege logs primarily involve the exchange of metadata fields, which are “used to describe the structural information of a file . . . as opposed to describing the content of a file.”³ The majority of these privilege logs are automatically generated and typically include fields like date, author, recipients, document type or file extension (g., .doc), and title or subject line (or file name). As a result, some courts refer to these logs as “Automated log[s].”⁴ That name is somewhat misleading because such logs generally also include a “privilege type” field populated by human document reviewers. These reviewers will also often draft descriptions for a “subject” field during their review. Still, this type of partial automation has become widespread over the last decade, receiving endorsements by the courts⁵ and industry publications alike.⁶
• Categorical privilege logs list related items by category, rather than by individual entry, and are typically organized by subject matter or communication type. Parties may agree that certain categories of communication—for example, communications with outside counsel after a lawsuit was filed—may be omitted from the privilege log.⁷ Because categorical logs provide for the exclusion of large batches of documents, they offer significant cost-saving opportunities. Defining categories can require substantial negotiation between parties, which of course cannot be automated. Once the parties define their categories, however, automation tools can group documents into categories according to predetermined parameters. This method of privilege logging can also be paired with other approaches to maximize efficiency.
• Phased privilege logs build upon categorical privilege logs to only include documents in particularly controversial categories as negotiated by counsel. Thus, these privilege logs currently provide little further opportunity for automation.

Advancing Automation and Other Cost-Saving Methods

Parties should continue to incorporate automation, machine learning, and other cost-saving methods to log privileged documents. In addition to the methods described above, some parties now automate nearly complete “draft” privilege logs that are then reviewed and revised by attorneys.⁸ Attorneys have also begun to use technology assisted review (“TAR”) to elevate likely privileged documents in the first instance. If attorneys adequately train the TAR models on the front end, this method could prove quite valuable for litigants. Future use cases may also include the use of generative AI models or other new technology.

In addition, courts increasingly have approved other techniques to reduce privilege logging burdens in cases with many documents. For example, parties now routinely prepare only one log entry for a single, wholly privileged email chain, rather than prepare a privilege log entry for every constituent email of such a chain.⁹ If, for some reason, the receiving party believes that the privilege log entry for that email chain is over-inclusive or not sufficiently descriptive, they could request that the producing party log each email in the chain separately in a supplemental privilege log. This type of provision allows parties to avoid wasting time logging documents that are of no interest to either side. Moreover, these provisions are now consistent with industry practice.¹⁰

Despite these trends, automated privilege logging will likely never fully eliminate the human element in privilege reviews. Using automation to combine and improve standard privilege logging methods, however, has and will continue to gain traction among litigants and approval by the courts.

¹See, e.g., The Sedona Conference, The Sedona Conference Commentary on Privilege Logs (Public Comment Version, Feb 2024), https://thesedonaconference.org/sites/default/files/publications/Commentary-on-Privilege-Logs-Public-Comment-February-2024.pdf [“Commentary on Privilege Logs”].

²See, e.g., ESI Order at §14, United States ex rel. Rebecca Miller v. Reckitt Benckiser Pharm. n/k/a Indivior Inc., No. 1:15-cv-00017-JPJ-PMS, (W.D. Va. Apr. 22, 2024), ECF No. 182-1.

³The Sedona Conference, The Sedona Conference Glossary: eDiscovery and Digital Information Management, Fifth Edition, 21 Sedona Conf J. 263, 337–38 (2020).

⁴See, e.g., ESI Order at §14, United States ex rel. Rebecca Miller, No. 1:15-cv-00017-JPJ-PMS, ECF No. 182-1; ESI Order at §5.K.1, Flip Phone Games, Inc. v. PLR Worldwide Sales Ltd., No. 2:23-cv-00139, (E.D. Tx. Dec. 5, 2023), ECF No. 49.

⁵See, e.g., Dkt. 49 ¶ 5.K.1, Flip Phone Games, No. 2:23-cv-00139.

⁶See “Commentary on Privilege Logs,” 4.

⁷See, e.g., ESI Order at §G.4, In re Delta Dental Antitrust Litig., No. 1:19-cv-06734, (N.D. Ill. Nov. 2, 2020), ECF No. 348.

⁸See “Commentary on Privilege Logs,” 19.

⁹See, e.g., ESI Order at §14(c), In Re: Realpage, Inc., Rental Software Antitrust Litig. (No. II), No. 3:23-md-3071, (M.D. Tenn. Feb. 16, 2024), ECF No. 815.

¹⁰See, e.g., ESI Order at §10(b), In re Diisocyanates Antitrust Litig., No. 2:18-mc-01001, (W.D. Pa. 2020), ECF No. 313; ESI Order at §G.4, In re Delta Dental Antitrust Litig., No. 1:19-cv-06734, (N.D. Ill. Nov. 2, 2020), ECF No. 348.