“Three-strata validation” was requested and rejected in In re Class Action Settlement Administration Litigation, 2026 WL 2444936 (D.D.C. Aug. 19, 2026). The parties in that case had narrowed their disputes and their “only remaining disagreement concerns how a party producing documents in discovery must validate that its productions are substantially complete.”

Plaintiffs asked for “three-strata validation.”  Defendants objected to the second and third strata.

The court explained the “three-strata” approach as:  “validation of documents coded as non-responsive, validation of documents coded as responsive, and validation of documents excluded from the universe of ESI material subject to discovery review.”

THE PARTIES AGREED TO NULL SET TESTING

The parties agreed that there is an obligation  “to confirm that responsive documents are not erroneously coded as non-responsive and omitted from their productions.” This involves review of a “statistically valid random sample” of documents coded as non-responsive. Id. at *1.

This standard practice is commonly called “elusion testing,” review of the “discard pile,” or, review of  the “null set.”  See Deal Genius, LLC v. O2COOL, LLC, 682 F. Supp. 3d 727, 734–35 (N.D. Ill. 2023) (“explaining that elusion testing, where the producing party reviews a random sample from the ‘null set’ of documents that did not hit on search terms and documents that were deemed non-responsive, is a ‘standard quality assurance practice’ that can ‘confirm the reasonableness and proportionality’ of the production”).

VALIDATION OF DOCUMENTS CODED AS RESPONSIVE – REJECTED

Defendants argued that validation of documents coded as responsive, to find false positives, was superfluous.  They asserted that Fed.R.Civ.P. 26(g) required that they certify that their response is complete and correct.  The court wrote: “Rephrased in statistical terms, producing parties must confirm that their false-negative rate falls within acceptable limits.”  In my less elegant words, a production may include nonresponsive documents; but, it cannot be a “dump truck.”

The court explained:

Review of documents marked responsive falls outside the bounds of litigants’ Rule 26(g) obligations because such review cannot identify whether documents are missing from a production, rendering it incomplete. Confirming that documents designated responsive are truly responsive reveals only the production’s false-positive rate, a proxy measure for whether the litigant has overproduced discovery material. But the Federal Rules do not prohibit parties from overproducing documents. Accordingly, this Court will not require defendants to incur additional costs re-reviewing documents that they have previously coded as responsive and intend to produce, barring claims of privilege. [emphasis in original].

VALIDATION OF DOCUMENTS EXCLUDED FROM THE UNIVERSE – REJECTED

The second strata consists of “ documents excluded from the universe of ESI material subject to discovery review.”  In other words, Plaintiffs asked for a review of documents that Defendants had not collected and that, presumably, were  not in Defendants’ litigation review platform.

The court wrote that examination of this set of documents, the “Excluded Set,” “could reveal false-negatives: responsive materials that should be produced but were erroneously omitted from the universe of materials collected for discovery review.”

However, the court responded that Plaintiffs’ request was duplicative and made too late, long after collection.  It wrote:  “Yet the Court is puzzled why plaintiffs request review of the Excluded Set after discovery review is largely complete, given that the parties also intend to review the Excluded Set for responsive material prior to beginning that review.” [emphasis in original].

The court explained that “ESI discovery generally proceeds in two phases: document collection and merits review.”  The parties had earlier “set forth agreed upon parameters for collecting the universe of documents that will be subject to merits review.”  That earlier agreement had required statistical sampling.

Insofar as the parties have agreed to statistically validate the completeness of the set of documents collected for merits review before even beginning that review, the Court finds that further validation of the Excluded Set after merits review is largely complete would be duplicative and unnecessary.

However, Plaintiffs made a different argument—so-called “substantial completeness evaluation.”  They argued that:

[T]heir proposed validation efforts “answer different questions at different points in time,” because search term validation “tests whether agreed terms are performing as intended before review begins,” while substantial completeness validation of the Excluded Set “tests whether the entire process — including all downstream coding decisions, review workflow, and production choices — resulted in a complete production at its conclusion.”

This argument was unsuccessful because the court held that “the Excluded Set is a collection of documents excluded before merits review begins and not subject to downstream coding decisions, review workflows, and production choices.”

Plaintiffs also contended that “the prevailing practice is for parties producing discovery to validate the Excluded Set both upon its creation and after merits review….”  The court noted that the authorities are split; however, it held that “the Court will not require producing parties to incur the cost and burden to re-validate the Excluded Set at the substantial completeness stage, provided that they statistically validate that documents excluded from merits review do not contain responsive material before beginning that review.”

I would add that “substantial completeness validation” of the Excluded Set could de-rail a case and mandate a “do over.”  If a sufficient number of false negatives was found in that set—and sufficiency is always debatable–presumably Defendants would have to engage in a new collection, review, and production process.

This blog was initially posted on  Electronic Discovery Reference Model

In Campbell v. TidalHealth, Inc., 2026 WL 2162749 (D. Md. Jul. 28, 2026)(Austin, J.), the Court wrote: “Campbell denies submitting documents into any generative artificial intelligence tools but admits to using unspecified ‘available tools solely to express [her] arguments in proper legal format, the same way any litigant uses available legal resources.’” Id. at *8.

After cautioning Campell about the use of GenAI, the Court wrote:

In addition, Campbell must ensure that any use of generative artificial intelligence does not violate the stipulated order governing confidential materials…. TidalHealth expresses concern that Campbell violated the February 27, 2026 Protective Order’s prohibition on submitting discovery material to public generative artificial intelligence tools. … see ECF 21 at ¶ 8 (imposing limitations on use of generative artificial intelligence). Campbell denies doing so…. If future filings suggest improper submission of discovery materials to generative artificial intelligence tools, the Court will be more inclined to issue a show cause order and consider sanctions.

Id. at *10 (emphasis added).

These orders are standard of care and are most often entered by consent. But not all orders are created equal and they are not always fair.  I have echoed Craig Ball’s novel post on the need for equity.  Another A.I. Protective Order – Part 10 quoting Craig Ball, The AI Protective Order Double Standard | Ball in your Court (Jul. 27, 2026).

UPDATE:

“Because the parties’ proposed protective order adequately complies with the standards set forth above, the Court will enter an order based on the parties’ submission. The Court makes four minor modifications to the parties’ protective order: … and (4) to establish parameters for the use of Artificial Intelligence as it pertains to confidential documents, see infra Paragraph (F).”  Zask v.Kohl’s Corp., 2026 WL 2531856 (E.D. Wisc. Aug. 27, 2026)(emphasis added). The order states:

F. USE OF ARTIFICIAL INTELLIGENCE (“AI”)
A receiving party may use AI tools in connection with discovery materials produced in this action only if the AI tool is operated in a secure environment and is subject to enforceable contractual terms that (a) prohibit the provider from retaining, using, disclosing, selling, or using the materials to train, improve, or develop any AI system; (b) prohibit the materials from being made publicly available or accessible to any other customer or third party; and (c) require commercially reasonable administrative, technical, and physical safeguards to protect the confidentiality of the materials. Discovery materials, whether designated CONFIDENTIAL or ATTORNEYS’ EYES ONLY, shall not be entered into any publicly available or consumer-based AI platform. Any use of AI tools must remain consistent with this Order and with the requirement that discovery materials be used solely for purposes of this litigation.

One defendant’s privilege log was untimely.  Plaintiff moved to compel, challenging privilege based on alleged subject matter waiver by production of privileged materials and also on an alleged untimely and deficient privilege log.

The court wrote that: “In diversity cases, the application of the attorney-client privilege is governed by state law—in this case, the law of South Carolina.”

As to the waiver issue, the court found that the communications between defendant’s general counsel and others were not for the primary purpose of securing legal advice.  Because they were not privileged, there was no subject matter waiver by disclosure.

Turning the choice of law issue regarding the privilege log, the court wrote:

As an initial matter, Stiles mistakes the extent to which South Carolina law governs the conflict over its privilege log. In Erie R.R. v. Tompkins, 304 U.S. 64, 58 (1938), the Supreme Court held that federal courts sitting in diversity must apply state substantive law. Under the Erie Doctrine, “a federal court is to apply state substantive law and federal procedural law in diversity cases.” …. Assertion of attorney-client privilege requires a party to satisfy both procedural and substantive criteria…. In claiming that the privilege applies, Stiles is correct that the Court’s assessment of whether attorney-client privilege applies to the documents at issue is governed by South Carolina law…. However, the production of a privilege log asserting attorney-client privilege over certain documents in order to withhold that information from discovery is procedural and, therefore, governed by federal law…. Accordingly, the Court applies federal law to determine whether Stiles’ privilege log was sufficient. [cleaned up; emphasis added].

The court held: “Upon review, the Court finds that these disclosures on Stiles’ privilege log are insufficient to allow Plaintiff to determine whether Stiles has properly asserted privilege over these documents.”

The court determined that it had four options: “(1) give the party another chance to submit a more detailed log; (2) deem the inadequate log a waiver of the privilege; (3) inspect in camera all of the withheld documents; and (4) inspect in camera a sample of the withheld documents.”  It engaged in detailed factual analysis and granted plaintiff’s motion in part and denied it in part.

This blog was initially posted on  Electronic Discovery Reference Model.

Basically, these orders limit use of artificial intelligence by the recipient of discovery materials.  But, they are not always fair.

Craig Ball has explained that these orders can create a double standard.  He quoted a District of Colorado decision stating “that practically speaking, and in light of the current state of AI, this provision will (at least for now) bar the parties from using most, if not all, mainstream low-to-no cost AI to process Confidential Information.”  Craig added:

If other courts follow uncritically, it will do what every prior technology-gatekeeping effort has done: widen the gap between well-funded litigants and everyone else, while delivering no meaningful improvement in data security.  My hope is that this post will shed light on a distinction without a difference so as to not hinder the use of properly configured, ‘consumer grade’ AI for processing sensitive data.

U.S. v. Dygdon, 2026 WL 2225226 (S.D.N.Y. Aug. 3, 2026), provides another example of an AI  protective order entered by consent in a criminal case.  The materials disclosed in discovery  would affect “the privacy, confidentiality and business interests of individuals and entities,” and “would risk prejudicial pretrial publicity if publicly disseminated….” The consent order states:

9. Restrictions on Use of Artificial Intelligence Tools to Process Disclosure Materials.

   a.  “AI tool” means any automated system that uses statistical modeling, machine learning, or similar techniques to process inputs and generate outputs, recommendations, or predictions, including but not limited to large language models, generative AI services, or AI-assisted software tools, whether cloud-based or otherwise.

   b.  Under no circumstances will any member of the defense team submit Disclosure Materials to a publicly accessible AI tool, or any AI tool, that retains and uses submitted data to train models. Such AI tools pose unique risks to the security and integrity of Disclosure Materials, given the practical inability to claw back or delete data once it has been incorporated into a model.

   c.  With respect to any AI tool used by the defendant, defense counsel must be able to certify that (i) the AI tool(s) will maintain the confidentiality of any Disclosure Materials; (ii) the AI tool(s) will not expose any Disclosure Materials to third parties not authorized under this Protective Order to receive such materials; (iii) the AI tool(s) will not retain or use any Disclosure Materials for model training; and (iv) all Disclosure Materials will be removed from the AI tool(s) at the conclusion of this case.

I suggest that there is general agreement that a degree of protection of materials produced in discovery is reasonable.  Craig has some excellent suggestions, writing: “I’m not arguing for anarchy or carelessness. I’m arguing for proportionality….”

Craig lists five provisions for a “properly scoped Ai provision….” They are: 1) no training; 2) no public accessibility; 3) matter isolation; 4) deletion at conclusion; and, 5) documentation.  He adds: “Five requirements, all achievable at any budget. All providing genuine protection against the actual risks that protective orders target: unauthorized use, competitive exploitation, and ongoing exposure. Anything beyond this isn’t really protecting data. It’s protecting market position.”

I suggest that AI protective orders are now the standard of care. I agree with Craig’s well-written suggestion that “we can protect discovery materials without building a toll booth that only the well-heeled can pass through.”

I also very much appreciate Craig’s endnote: “Hat tip to my friend Michael Berman, whose frequent and excellent series of posts about AI and discovery law got me thinking about this today.”

For more on this topic, please visit:

• Restrictions on Use of Artificial Intelligence Tools to Process Disclosure Materials (Jul. 27, 2026);
• Another A.I. Protective Order (Jul. 9, 2026);
•  A.I. Protective Order—Again (Jun. 18, 2026);
• Unicorn Rejects A.I. Protective/Confidentiality Order – Order Entered in Criminal Case (Jun. 4, 2026);
• Are Protective Orders Limiting Upload to AI of Documents Produced by an Opponent in Discovery Becoming the Standard of Care?  (May 26, 2026), discussing Rudasill v. Swiss Re American Holding Corporation, 2026 WL 1329451 (S.D.N.Y. May 13, 2026);
• Another Confidentiality Order Limiting Use of AI to Analyze Protected Materials (Apr. 27, 2026), discussing In re: ByHeart, Inc., Infant Formula Marketing, Sales Practices, and Products Liability Litigation,2026 WL 1103170 (S.D.N.Y. Apr. 23, 2026);
• A.I. Protective Orders Are Becoming Routine (Apr. 20, 2026), discussing Stansfield v. International Business Machines Corporation, 2026 WL 1031230, at *2 (D. Nev. Apr. 16, 2026);
• “Deepfake Executives, Created Via Depositions, Pose Grave Threat”: Protective Order Suggested (Apr. 5, 2026);
• Morgan v. V2X, Inc.,2026 WL 864223 (D. Col. Mar. 30, 2026), as also discussed in Important A.I. Work Product and Protective Order Decision (Mar. 31, 2026);
• Order Prohibiting Upload of Confidential Discovery Documents to Artificial Intelligence (“AI”) (Nov. 3, 2025), based on Warner v. Gilbarco, Inc., 2025 WL 3047881 (E.D. Mich. Oct. 30, 2025);
• Protective Order Limited Uploading Discovery Responses to Open A.I.(Mar. 26, 2026), discussing Jeffries v. Harcros Chemicals, Inc.,2026 WL 820218 (D. Kans. Mar. 25, 2026).

This blog was initially posted on  Electronic Discovery Reference Model.

UPDATE:

After the first posting of this blog, an AI protective order was entered by consent in U.S.A. v. Marrow, 2026 WL 2389395 (S.D.N.Y. Aug. 17, 2026).

7. Restrictions on Use of Artificial Intelligence Tools to Process Disclosure Materials.

a. “AI tool” means any automated system that uses machine learning, or similar techniques to process inputs and generate outputs, recommendations, or predictions, including but not limited to large language models, generative AI services, or AI-assisted software tools, whether cloud-based or otherwise.

b. The defense team will not submit Disclosure Materials to a publicly accessible version of an AI tool (e.g., ChatGPT, Claude, Grok, Gemini, Copilot, Perplexity, Midjourney, Canva). The defense team will not submit Disclosure Materials to an AI tool for which the AI tool provider or vendor (e.g., OpenAI, Anthropic, Microsoft, Google) uses submitted data for its own model training, allows submitted data to be publicly accessed, discloses submitted data to third parties, or, except as specified below, retains submitted data.

c. The defense may submit Disclosure Materials to an AI tool (i) that maintains the confidentiality of Disclosure Materials; (ii) that will not expose any Disclosure Materials to third parties not authorized under this Protective Order to receive such materials; (iii) that will not retain or use any Disclosure Materials for model training; and (iv) all Disclosure Materials will be removed from the AI tool(s) at the conclusion of this case.

Additional consent cases are U.S.A. v. Spatola, 2026 WL 2351459 (S.D.N.Y. Aug. 13, 2026), U.S.A. v. Davis, 2026 WL 1333520 (S.D.N.Y. Aug. 12, 2026).  The protective order in Spatola is the same as Marrow.  The order in Davis states:

Restrictions on Use of Artificial Intelligence Tools to Process Disclosure Materials.

a.  “AI tool” means any automated system that uses statistical modeling, machine learning, or similar techniques to process inputs and generate outputs, recommendations, or predictions, including but not limited to large language models, generative AI services, or AI-assisted software tools, whether cloud-based or otherwise.

b. Under no circumstances will any member of the defense team submit Disclosure Materials to a publicly accessible AI tool, or any AI tool, that retains and uses submitted data to train models. Such AI tools pose unique risks to the security and integrity of Disclosure Materials, given the practical inability to claw back or delete data once it has been incorporated into a model.

A different A.I. order was entered by consent in U.S. v. Stamp, 2026 WL 2321897 (S.D.N.Y.  Aug. 11, 2026).  It added:

d. Any translation, transcript, summary, or other output that reveals the substance of Disclosure Material shall be treated as Disclosure Material, but draft legal briefs, counsel memoranda, and other attorney work product shall not be considered Disclosure Material irrespective of whether they were created in part through the use of an AI tool.

e. Defense counsel may permit access by AI tool provider personnel, affiliates, and subprocessors, to databases containing Disclosure Material only as reasonably necessary to provide, maintain, secure, or support the AI tool, subject to confidentiality obligations set forth herein.

The issue before the court arose out of the developer’s motion to compel. Blue Moon made an interesting challenge to privilege based on a disclosure to the condominium’s management agent.

This blog, however, addresses the search methodology issue and Blue Moon’s assertion that Trailsedge must search all current and former unit owners’ email and text messages.  Trailsedge Owners Ass’n., Inc. v. NED Ski Associates, LLC, 2026 WL 2083546 (D. Vt. Jul. 16, 2026).

While the court rejected defendants’ untimely demand to use their search terms, it found gaps in plaintiff’s keywords and ordered broad disclosure.  The court rejected defendants’ demand to search the ESI of all current and former unit owners over more than a decade.  However, based on certain statements of plaintiff, it found that plaintiff had opened the door to additional unit owner disclosure, which was ordered by the court.

SEARCH METHODLOLGY

In my opinion, both parties were at fault when it came to search terms.  Trailsedge decided on its keywords, unilaterally and without consultation.  However, “Trailsedge argues that Blue Moon … never asked to negotiate search terms, and failed to propose its own terms until approximately two years later.”  Id. at *3.

But, “[i]n reply, Blue Moon denies it acquiesced to Trailsedge’s unilateral search process, arguing it did not know Trailsedge had used keyword search terms until many months after the discovery production and was entitled to assume Trailsedge had produced all responsive materials…. Blue Moon also argues that the scope of the requested search is proportional because Trailsedge seeks damages exceeding $10 million for alleged defects dating back to the period when the buildings were constructed and occupied while refusing to identify the date range, custodians, and results of its searches.”  Id. at *4.

The court looked to the Sedona Conference, id. at *4, passim, and  Judge Peck’s decision in William A. Gross Constr. Assocs., Inc. v. Am. Mfrs. Mut. Ins. Co., 256 F.R.D. 134 (S.D.N.Y. 2009), writing that “when a party uses keyword searches to identify responsive ESI, the search process must be cooperative, well-informed, capable of explanation, and sufficiently transparent to permit the opposing party and the court to evaluate it if challenged.”  It wrote that cooperation and proportionality are the dominant drivers. “As Gross explained, keyword searches must be developed with input from custodians who know the terminology used in the relevant communications and must be tested to ensure accurate retrieval and reduce false positives.” Id. at *5.

The Trailsedge court continued:  “Applying the principles described above to this case, the court first rejects Blue Moon’s request to impose its proposed keyword search” on Trailsedge.  Id. at *5.  It found that those terms may be overbroad, redundant, “or likely to generate excessive false positives.”  Further, Blue Moon was untimely:

Sedona’s principles do not give a party license to wait until late in discovery and then demand a wholesale redo of its opposing party’s collection process using its preferred terms….  Trailsedge also notes that Blue Moon did not request search-term information or propose its own search terms until approximately two years later…. The timing and proportionality considerations weigh against granting Blue Moon’s request in full.

Id. at *5.

However, Trailsedge was not off the hook with its unilateral selection of keywords. It wrote that: “The court… also rejects Trailsedge’s argument that Blue Moon’s delay ends the inquiry. Trailsedge chose to use keyword searches as part of its collection process, and the reasonableness of that process became a fair subject of discovery once it was challenged.” Id. at *6.

Again citing Sedona and Judge Peck, the Trailsedge court noted “the more basic principle that counsel must understand, supervise, and be able to describe the search process used to identify responsive ESI in enough detail to assess its adequacy.” Id. at *6.  Finding “apparent gaps” in the terms designed unilaterally by Trailsedge, the court wrote:

Trailsedge has not met its obligation to provide enough information to permit a meaningful assessment of whether the searches were reasonable and proportional. The record does not show that Trailsedge has identified the custodians searched, the date ranges applied, the dates the searches were conducted, the terms used for each custodian or source, or the results returned by each term….  Trailsedge must therefore provide additional information about its search methodology.

Id. at *6.  It added that Trailsedge’s  “omission of central party names and words related to critical legal issues raises a sufficient question about the adequacy of Trailsedge’s prior search sufficient to warrant additional discovery.”  Id.

Accordingly, within fourteen days, Trailsedge shall serve on Defendants a written description of its prior ESI search methodology. That description shall identify: (1) each custodian whose emails, text messages, or other ESI were searched; (2) each data source searched; (3) the date range applied to each search; (4) the date on which each search was performed; (5) the search terms used for each custodian or source; and (6) the number of hits returned by each term (to the extent reasonably available). Trailsedge shall also state whether custodians were instructed to use particular search terms, whether custodians self-selected additional terms, and what steps counsel took to supervise or validate the searches.

Id. at *6.  This was to be followed by a court-ordered “meet and confer.” Id.; see generally, the Hon. Paul W. Grimm (ret.), Michael D. Berman, et al., “Discovery about Discovery: Does the Attorney Client Privilege Protect All Attorney-Client Communications Relating to the Preservation of Potentially Relevant Information,” 37 U. Balt. L. Rev. 413 (2008).

SEARCHES OF EVERY CURRENT AND FORMER UNIT OWNER

The court explained the unit owner issue as follows:

Blue Moon seeks an order requiring Trailsedge to collect and produce responsive documents from all current and former unit owners for the period of January 1, 2011 through May 12, 2023…. Blue Moon argues Trailsedge was required to search all unit owners’ files because Trailsedge “is suing on behalf of all unit owners” and because Trailsedge’s initial disclosures allegedly identified members, not just Board members, as persons likely to have discoverable information.

Id. at *7. “Trailsedge responds that the Association is a Vermont nonprofit corporation distinct from its members and that the unit owners are not parties or custodians of Association records merely because they own units…. Trailsedge further argues that, except in limited governance matters, the Association acts through its executive board under the Vermont Nonprofit Corporation Act and that the Association’s statutory authority to litigate in its own name on behalf of itself or unit owners does not collapse the distinction between the Association and its members….  Trailsedge also argues that the request is disproportionate because Trailsedge consists of twenty-four units that have been bought and sold over the years by an unknown number of owners and Blue Moon seeks documents from January 1, 2011 through May 12, 2023.”  Id. (emphasis added).

The court resolved the issue by applying the “possession, custody, and control” and proportionality  standard of Fed.R.Civ.P. 34.  “In the District of Vermont, ‘control’ is construed broadly and is not limited to legal ownership or actual physical possession; documents may be within a party’s control when the party has the ‘right, authority, or practical ability’ to obtain them.”  Id. at *7.

It held that Blue Moon had the burden of proving control by Trailsedge and failed to carry it: “The record also does not support treating all unit owners as custodians simply because some unit owners may possess relevant information. Rule 34 control requires more than relevance.”  Id. at *7.  The court noted that the Sedona conference calls for “caution” in evaluating information in the possession of non-parties.

Although the [Sedona Conference] Commentary’s recommended “legal right” standard is narrower than the “practical ability” standard applied by district courts in the Second Circuit, it identifies generally applicable practical and fairness concerns implicated when discovery obligations are extended to ESI held by third parties. The Commentary explains that an expansive approach may require parties to preserve, search, and produce documents they do not actually possess and may not have the ability or legal right to obtain…. That concern is present here. Blue Moon has shown that unit-owner communications may be relevant, but it has not shown that Trailsedge has actual possession, legal authority, or a practical ability to obtain every current and former unit owner’s personal emails and text messages on demand.

Id. at *7 (emphasis added).  The court also evaluated proportionality: “The documents Blue Moon seeks may be relevant to notice, knowledge, and statute-of-limitations issues. Blue Moon’s requested owner-wide search, however, would encompass more than twelve years, cover all current and former owners of twenty-four units, and include potentially sensitive personal texts and emails.”  Id. at *8.

However, Trailsedge was not off the hook. Its discovery responses “reasonably could have led Blue Moon to believe that at least some non-Board Association members had supplied documents or information.” Id. at *8.  In my words, Trailsedge opened the door.  Therefore:

In light of the ambiguity created by Trailsedge’s interrogatory responses, however, the court finds the responses require clarification. Rule 26(g) requires Trailsedge’s discovery responses to be formed after a reasonable inquiry, and Rule 26(e) requires supplementation if the responses were materially incomplete or incorrect. Fed. R. Civ. P. 26(e), (g). Because Trailsedge’s responses suggest that some member-produced materials were considered but do not identify which members provided documents, what sources were searched, or whether any non-Board member communications were requested or reviewed, Trailsedge must clarify those issues in the search-methodology disclosure ordered in the previous section. The disclosure shall state whether Trailsedge requested documents from any non-Board unit owners; identify any non-Board unit owners from whom Trailsedge obtained documents or information used in preparing its discovery responses or productions; identify the categories of documents obtained from those members; and certify whether Trailsedge has produced all responsive, nonprivileged documents from those member-provided materials that are within its possession, custody, or control.

Id. at *8.

For more on “possession, custody, and control,” see:

• Possession, Custody, or Control – Need for a Uniform National Standard – Part II;
• Isn’t It Time for a Uniform National Standard on “Possession, Custody, or Control”?

They sought depositions in excess of those provided for by the Federal Rules of Civil Procedure based on documents that had been produced from multiple custodians.  The court held that not every document custodian should be deposed and: “Custodians are chosen because their files are likely to contain responsive documents, which is a reason to collect their files, not necessarily a reason to depose them.”  Id. at *3.

Plaintiffs asked for leave to take 27 fact depositions of party witnesses, “without prejudice to seeking more.”  In response:

LinkedIn asks the court to (1) hold the plaintiffs to the presumptive ten depositions, (2) require them to serve their Rule 30(b)(6) notice now, and (3) require them to coordinate with LinkedIn on overlapping Rule 30(b)(1) and 30(b)(6) witnesses.

The ruling moved toward the center: “ The court grants the plaintiffs’ motion in part: the plaintiffs may take up to fifteen depositions of party witnesses, a number that includes one Rule 30(b)(6) deposition of LinkedIn, without prejudice to a renewed, witness-specific showing of need for additional depositions.”  They were directed to coordinate overlapping depositions.

The court wrote that: “Document discovery to date has been extensive: LinkedIn has nineteen designated custodians, including former CEO Ryan Roslansky and current CEO Daniel Shapero, and it has produced thousands of custodial documents.”

The 27 requested depositions were party witnesses. They included current or former senior executives and four groups of current or former employees. The court wrote: “For each proposed deponent, the plaintiffs cite produced documents that, they contend, show the witness’s personal involvement in the challenged conduct (which spans ten years) or unique knowledge of market definition, monopoly power, or pricing, all topics central to the plaintiffs’ claims.”

“LinkedIn responds that the plaintiffs’ showing of particularized need largely rests on the witnesses being designated custodians or appearing in the document production. It contends that the plaintiffs misread many of the documents they cite and provides an appendix that, it asserts, provides the fuller context for those produced materials. LinkedIn contends that several of the plaintiffs’ proposed deponents (including an in-house lawyer and the engineer who leads LinkedIn’s AI-inference platform) have nothing to do with the challenged conduct, and that other witnesses are at least facially duplicative because they work on the same issues and teams.”

The court explained that: “Rule 30(a)(2) limits the number of depositions that each side may take (without leave or stipulation) to ten. A party seeking to exceed the presumptive number of depositions must make a particularized showing of the need for the additional discovery…. To that end, courts will generally not grant leave to expand the number of depositions until the moving party has exhausted the ten depositions permitted as of right under Rule 30(a)(2)….. When evaluating a party’s request to expand discovery limitations, the court must ensure that the discovery sought is relevant to the party’s claims or defenses and not unreasonably cumulative, duplicative, or burdensome.”  Id. at *2 (cleaned up).

The court found that 10 depositions were too few: “The plaintiffs’ claims are ‘narrow’ in the sense that they target two categories of conduct, but that conduct spans more than a decade, hundreds of API agreements, a major cloud migration, and pricing across multiple Premium product lines. The document production to date reflects this scope, with thousands of custodial files produced. And this court’s previous orders have acknowledged the breadth of the underlying conduct by ordering additional custodians and search terms. While sometimes courts require a party to exhaust the depositions it is permitted before seeking more, courts depart from this when there are multiple parties and the case’s complexity warrants more depositions.”  Id. at *3. That situation was presented here. Id.

“But,” the court continued, “twenty-seven depositions is too many, at least on this record. Much of the plaintiffs’ showing hinges on the requested witnesses being designated custodians or appearing on produced documents. This falls short of the plaintiffs’ burden of making a particularized showing of need: custodial status and document counts do not, by themselves, establish that a witness will offer relevant and non-cumulative deposition testimony.”  Id. at *3 (emphasis added).  In short, not every custodian must be deposed. Id.  The court also identified a “duplication problem” in plaintiffs’ request.  “And for some of the twenty-seven witnesses, relevance is thin….” Id.

The court’s holding was: “A reasonable approach lies between the parties’ positions: leave to take up to fifteen depositions of party witnesses, a number that includes one Rule 30(b)(6) deposition of LinkedIn (which counts as one deposition regardless of the number of designees)….  This middle course is consistent with decisions managing comparable cases.”  Id. at *4.  It left the door open to additional, particularized requests.

This blog was initially posted on  Electronic Discovery Reference Model.

Restrictions on Use of Artificial Intelligence Tools to Process Disclosure Materials.

a.  “AI tool” means any automated system that uses statistical modeling, machine learning, or similar techniques to process inputs and generate outputs, recommendations, or predictions, including but not limited to large language models, generative AI services, or AI-assisted software tools, whether cloud-based or otherwise.

b.  Under no circumstances will any member of the defense team submit Disclosure Material to a publicly accessible AI tool, or any AI tool, that retains and uses submitted data to train models. Such AI tools pose unique risks to the security and integrity of Disclosure Material, given the practical inability to claw back or delete data once it has been incorporated into a model.

c.  With respect to any AI tool used by the defendant, defense counsel must be able to certify that (i) the AI tool will maintain the confidentiality of any Disclosure Material; (ii) the AI tool will not expose any Disclosure Material to third parties not authorized under this Protective Order to receive such materials; (iii) the AI tool will not retain or use any Disclosure Material for model training; and (iv) all Disclosure Material will be removed from the AI tool at the conclusion of this case.

U.S.A. v. Mora, 2026 WL 2058416 (S.D.N.Y. Jul. 16, 2026).

This is no longer front page news. It is becoming standard practice.

One provision appears problematical to me—”the AI tool will not expose any Disclosure Material to third parties not authorized under this Protective Order to receive such materials.”  Given most A.I. privacy policies, that seems an excessively high standard.

One hypothetical discussed in some of the following blogs and reflected in this order, is the problem of exercising “clawback” rights if material has been uploaded to A.I.

Assume that A and B have a Fed.R.Evid. 502 non-waiver order and a Fed.R.Civ.P. 26 “clawback” agreement. Assume further that A produces a privileged document to B in discovery.  The next day, B uploads it to consumer-grade AI.  Assume that the following day, A exercises its “clawback” rights. How does B return, sequester, or destroy the document on A.I.?

For more information on related orders, please see:

• Another A.I. Protective Order (Jul. 9, 2026);
•  A.I. Protective Order—Again (Jun. 18, 2026);
• Unicorn Rejects A.I. Protective/Confidentiality Order – Order Entered in Criminal Case (Jun. 4, 2026);
• Are Protective Orders Limiting Upload to AI of Documents Produced by an Opponent in Discovery Becoming the Standard of Care?  (May 26, 2026), discussing Rudasill v. Swiss Re American Holding Corporation, 2026 WL 1329451 (S.D.N.Y. May 13, 2026);
• Another Confidentiality Order Limiting Use of AI to Analyze Protected Materials (Apr. 27, 2026), discussing In re: ByHeart, Inc., Infant Formula Marketing, Sales Practices, and Products Liability Litigation,2026 WL 1103170 (S.D.N.Y. Apr. 23, 2026);
• A.I. Protective Orders Are Becoming Routine (Apr. 20, 2026), discussing Stansfield v. International Business Machines Corporation, 2026 WL 1031230, at *2 (D. Nev. Apr. 16, 2026);
• “Deepfake Executives, Created Via Depositions, Pose Grave Threat”: Protective Order Suggested (Apr. 5, 2026);
• Morgan v. V2X, Inc.,2026 WL 864223 (D. Col. Mar. 30, 2026), as also discussed in Important A.I. Work Product and Protective Order Decision (Mar. 31, 2026);
• Order Prohibiting Upload of Confidential Discovery Documents to Artificial Intelligence (“AI”) (Nov. 3, 2025), based on Warner v. Gilbarco, Inc., 2025 WL 3047881 (E.D. Mich. Oct. 30, 2025);
• Protective Order Limited Uploading Discovery Responses to Open A.I.(Mar. 26, 2026), discussing Jeffries v. Harcros Chemicals, Inc.,2026 WL 820218 (D. Kans. Mar. 25, 2026).

This blog was initially posted on  Electronic Discovery Reference Model.

1. The Parties agree that each Responding Party is best situated to determine the most appropriate method or methods for that Responding Party to search, collect, cull, and produce documents responsive to discovery, consistent with Sedona Conference Principle 6 which instructs that “[r]esponding parties are best situated to evaluate the procedures, methodologies, and technologies appropriate for preserving and producing their own electronically stored information.” Nothing in this ESI Protocol alters a Responding Party’s obligation to conduct a reasonable inquiry or use a reasonable process in searching for and producing relevant information, nor does it alter a Requesting Party’s right to challenge a deficient production.
2. The Parties will meet and confer to discuss the use of search terms, filters and date ranges or the use of advanced search and retrieval technologies. Within a reasonable time period, Responding Party shall specify the technique or techniques (e.g., search terms, technology assisted review (“TAR”), artificial intelligence or machine-learning technology (collectively, “AI”)) it will use to search for and cull material it reasonably anticipates will be responsive to Requests for Production. If the Requesting Party objects to the particular technique or techniques, the Parties will timely meet and confer regarding the matter and submit any unresolved disputes to the Court for resolution.
3. A Party who intends to use search terms will describe a Search Protocol that will include: (a) the criteria to be used to identify the universe of Documents to which search terms will be applied (e.g., date range), and (b) proposed search terms to be applied to that universe of Documents, subject to revision based on, for example, meet and confer of the Parties and the number of hits returned by each term. A Requesting Party may also suggest search terms to be applied. Each Requesting Party shall limit its ESI production requests to a total of twenty search terms per custodian per party, for the avoidance of doubt a custodian can include either Party itself for searches of Party-wide electronic repositories. The Parties may jointly agree to modify this limit without the Court’s leave. The Court shall consider contested requests for additional search terms per custodian, upon showing a distinct need based on the size, complexity, and issues of this specific case. The search terms shall be narrowly tailored to particular issues. Indiscriminate terms, such as the producing company’s name or its product name, are inappropriate unless combined with narrowing search criteria that sufficiently reduce the risk of overproduction. The Parties acknowledge that “indiscriminate terms” would not include names of various publicly available training datasets such as, “Books3.” A conjunctive combination of multiple words or phrases (e.g., “computer” and “system”) narrows the search and shall count as a single search term. A disjunctive combination of multiple words or phrases (e.g., “computer” or “system”) broadens the search, and thus each word or phrase shall count as a separate search term unless they are variants of the same word or identify alternative names, examples, or members of the same discrete set of datasets, websites, or repositories. Use of narrowing search criteria (e.g., “and,” “but not,” “w/x”) is encouraged to limit the production.
4. Any Search Protocol will include a requirement for the Responding Party to provide hit reports for proposed search terms prior to utilizing the search terms to narrow the universe of Documents to be reviewed. Search term hit reports will include, for each term, the number of documents with one or more search term hits, that number plus any attachments to those documents, and the number of documents hitting uniquely on that term and not on any other term in the same list. The report will also include total number of documents hitting on one or more terms and that number plus any attachments to those documents (presumptive review population). Any Search Protocol shall include a requirement to review one random statistical sample of distinct Documents, determined by a sampling of a confidence level of 95% and a margin of error of 2%, that do not hit on the search terms that are agreed upon by the Parties (the “Null Set”). The Null Set of distinct Documents shall be generated after all document processing is substantially complete. These Null Set Documents should come from each custodian and include Documents throughout the entire date range. The Responding Party should then review the Null Set Documents and include in future productions any responsive Documents located therein. If more than 10% of the Null Set are responsive, then the Parties shall revisit the agreed-upon search terms by analyzing the Documents produced and determining which additional terms are necessary. The Parties agree to meet and confer regarding the Search Protocol and to raise any disputes regarding the Search Protocol for resolution by the Court.
5. Nothing in this order or any other document relating to discovery in this matter shall be construed or interpreted as precluding a Responding Party from performing a responsiveness review to determine if documents captured by search terms are in fact responsive to the requesting party’s request. Further, nothing in this order or any other document relating to discovery in this matter shall be construed or interpreted as requiring the production of all documents captured by any search term if that document is – in good faith – deemed not responsive to the requesting party’s request for production by the Responding Party.
6. If a Responding Party elects to use search terms in combination with TAR on the same set of Documents (“layering”), it shall disclose that intent to the Requesting Party before doing so, and the Parties shall meet and confer on a protocol governing the use of layering. The protocol shall include a method to compare the number of hits when there is and is not layering for the set of Documents at issue, and permit the Requesting Party to make a reasonable request that the Responding Party review a reasonable sample of Documents excluded from review through the layering process to assess whether responsive Documents have been excluded. In the event that the Parties cannot reach an agreement on a protocol after meeting and conferring, the Parties shall submit competing protocols to the Court.
7. A Party who intends to use TAR to reduce the population of documents to be reviewed shall disclose a TAR Protocol that includes, at a minimum, the information listed in Appendix 3.
8. A Responding Party that elects to use AI to assist in responsiveness or privilege determinations (“AI Responsiveness Review”) shall disclose that election to the Requesting Party. AI Responsiveness Review constitutes workflows utilizing large-language models (LLMs), deep-learning classifiers, embedding-based similarity analysis, semantic clustering, predictive redaction systems, or generative summarization models used to determine whether a Document will be produced, withheld, or redacted.
9. The Responding Party shall provide an AI Review Protocol, which shall include, at a minimum, the information listed in Appendix 4.

Nothing in this Section shall be construed to diminish the Responding Party’s obligations under the Federal Rules of Civil Procedure, this Order, or applicable law. Legal counsel remains responsible for ensuring the accuracy and completeness of all productions.

Documents or ESI known to be responsive to a discovery request or relevant to the subject matter of this action shall be produced without regard to whether it was responsive to any search methodology described herein or developed in accordance with this Order, unless Counsel specifically identifies the documents being withheld and provides a specific objection for withholding each withheld document.

Documents which are reasonably believed to be responsive and for which text-based search technologies are fundamentally ineffective, such as images, video, certain spreadsheets, certain hard copy documents, certain documents from noncustodial sources, or certain foreign language documents where the Parties do not have suitable search terms in such language, must be reviewed without culling by search terms, predictive coding, or other technologies that rely primarily on text within the document. Prior to the production of such unsearchable items, the Responding Party may conduct a page-by-page review for responsiveness, confidentiality, privilege, and other protections.

In accord with Wilson Aerospace, it has long been recognized that a party’s duty to produce does not end with running search terms provided by an opponent.  Is There a Duty to “’Fess Up?” – Part II (Nov. 12, 2022).  “[A]n ESI protocol and search terms work in tandem with the parties’ obligations under the Federal Rules and do not replace a party’s independent obligation to produce electronic (or paper) documents that are reasonably accessible, relevant, and responsive within the meaning of Rule 34.”  Raine Grp. LLC v. Reign Capital, LLC, 2022 WL 538336, at *2 (S.D.N.Y. Feb. 22, 2022).  The protocol is an aid to each party fulfilling its independent obligation; it is not a replacement for that duty. Id.  Each party must sign and certify that it conducted a reasonable search. Id. at *2.

The Wilson Aerospace court’s starting point was to reiterate that Fed.R.Civ.P. 34 requires that a party must initiate a reasonable inquiry and produce responsive documents that are in its possession, custody or control.  It stated that this is an “affirmative duty.” Id. at *2.

Importantly, it added: “In matters involving electronically stored information, parties may employ custodian-based collections, agreed search terms, and other tools referenced in their ESI protocol to implement their Rule 34 obligations. But these tools do not narrow or override the Federal Rules.” Id. (emphasis added).

As a belt and suspenders argument, the court noted that the parties’ ESI Protocol confirmed that principle. It quoted that agreement:

Nothing in this Order shall preclude the producing party from producing additional relevant and responsive materials beyond those identified pursuant to the procedure described herein. Where appropriate, relevant and responsive non-privileged documents and ESI that are easily identifiable and segregable shall be collected and produced without the use of the procedure described herein. The discovery requests (in accordance with and subject to any limits or objections supported by the Federal Rules of Civil Procedure, the Local Rules, and any relevant orders of the Court) shall govern the scope of documents and ESI to be produced, subject to any agreements reached during the parties’ conferral, and otherwise search terms do not supplant discovery requests. The parties shall produce any relevant and responsive, non-privileged documents, to the extent they exist and can be located after a reasonably diligent search of relevant custodians and shared platforms, upon which the parties shall meet and confer.

Id. (emphasis added).

The Wilson Aerospace court then summarized the parties’ agreement: “Thus, the Protocol itself makes clear that: (1) discovery requests—constrained by the Federal Rules, local rules, and court orders—govern the scope of documents and ESI to be produced; (2) search terms and similar procedures do not supplant that scope; and (3) parties remain obligated to conduct a reasonably diligent search of relevant custodians and shared platforms and to produce relevant, responsive, non-privileged documents they can locate.”

The court wrote that: “The Protocol is a mechanism to implement the Rules, not a device to contract around them….  A party therefore cannot justify failing to collect obviously relevant documents on the ground that the relevant employee was not designated as an ‘ESI custodian’ or that such documents might fall outside an agreed search-term workflow.”  Id. at *2 (emphasis added).

Applying those principles to the facts of Wilson Aerospace: “Here, Defendant admitted that it did not search Patrick Murphy’s documents because he was not a disclosed custodian. Given Murphy’s apparent relevance, that position is inconsistent both with Defendant’s obligations under Rule 34 and with the plain language of the ESI Protocol quoted above. Defendant’s discovery duties are defined first and foremost by the Federal Rules and the scope of the discovery requests; the ESI Protocol cannot be invoked to avoid collecting and reviewing responsive documents from plainly relevant employees within Defendant’s control.”  Id. at *3 (emphasis added).

I suggest that there is a complicating factor and much turns on the parties’ agreements. Fed.R.Civ.P. 29(b) permits the parties to stipulate that depositions and “other procedures governing or limiting discovery be modified….”  [emphasis added].

The Advisory Committee Notes to the 1970 Amendment state that, in 1970, there was “no provision for stipulations varying the procedures by which methods of discovery other than depositions are governed.”  The Amendment “recognizes” the common practice of party agreements.

The 1993 Amendment revised the Rule “to give greater opportunity for litigants to agree upon modifications to the procedures governing discovery or to limitations upon discovery. Counsel are encouraged to agree on less expensive and time-consuming methods to obtain information, as through voluntary exchange of documents, use of interviews in lieu of depositions, etc.”  Further, the Advisory Committee Notes state that: “Under the revised rule, the litigants ordinarily are not required to obtain the court’s approval of these stipulations.”

I suggest that an ESI Protocol is in substance, if not in form, such a stipulation. To the same effect, one court recently wrote that: “The purpose of an ESI Protocol is to encourage reasonable electronic discovery with the goal of limiting the cost, burden and time spent, while ensuring that information subject to discovery is preserved and produced to allow for fair adjudication of the merits.” The Purpose of an ESI Protocol (Mar. 5, 2026).

One countervailing ambiguity arises because the Federal Rules of Civil Procedure do not recognize an “ESI Protocol.”   In fact, no Rule mentions an ESI Protocol. An “ESI Protocol” is Not a Rule 26(f) “Discovery Plan” (Nov. 24, 2025); “ESI Protocol” v. “Discovery Plan” (Jan. 2, 2024).

I fully agree with Wilson Aerospace to the extent that it emphasizes a rules-based duty. And, the court’s application to an obviously relevant custodian seems reasonable.  It does not appear that there was a meeting of the minds to exclude this custodian.

I wonder, however, whether the protocol’s language “subject to any agreements reached during the parties’ conferral,” might be inconsistent.  And, if it is their intent to do so, parties should be permitted to stipulate that certain witnesses be included or excluded.  Mutual intent is what seems to have been missing here.

This blog was initially posted on  Electronic Discovery Reference Model.

Clearly the court is privy to information I do not have and may have explained its reasoning in greater depth elsewhere. For example, the opinion opens by stating: “For the reasons discussed during today’s discovery conference, the parties’ current discovery disputes are resolved as specified below.”

ORDER THAT EVERY “HIT” MUST BE PRODUCED, UNLESS PRIVILEGED

First, the court ordered plaintiff Sussman to run defendant VWR’s two disputed search strings across all of its custodians:

Documents from October 1, 2018 to the present: Whiteboard AND NOT (*@whiteboardadvisors.com OR “whiteboard advisors” OR Edmentum OR “Paul Joy” OR “GSV’s News” OR “@morningbrew.com” OR “Carr (meeting rooms)”)

Documents from January 1, 2015 to the present: (Dwayne OR Dwayne’s) OR dwaynej*

It then ordered that “Sussman must produce the (i) hit reports for the searches specified above and (ii) all of the documents that hit on the search terms, except those (if any) withheld for privilege.” Id. at *1 (emphasis in original).

The court added: “For avoidance of doubt: Sussman may not withhold documents hit on by the two modified search strings set forth above based on a responsiveness review.”   Id. (emphasis added).

It may be that the court had sufficient information to determine responsiveness based on the detailed search string.

But I support the general rule that the fact that a document “hits” on a search term or string does not mean that it is responsive.  And, Fed.R.Civ.P. 26(b) limits the scope of discovery to information related to claims and defenses.  Non-responsive information fails to meet that standard.

For additional authorities, please see:

• Does Every “Hit” on a “Hit Report” Have to be Produced? (Nov. 13, 2024);
• How to Avoid Contentious “Hit Report” Problems – Part III (Nov. 15, 2023);
• How to Avoid Contentious “Hit Report” Problems (Jan. 4, 2023);
• Hit Reports (Aug. 14, 2022).

Best practice would be for the parties to cooperate and agree in advance.  Craig Ball posted Federal Court Rules on Whether Documents Containing Agreed-Upon Keywords are Responsive Per Se | Ball in your Court (Oct, 25, 2021). He wrote:

If you enter into an agreement with the other side to use keywords and queries for search, be clear about expectations with respect to the disposition of items hit by queries.  Assuming the items aren’t privileged, are they deemed responsive because they met the criteria used for search or is the producing party permitted or obliged to further cull for responsiveness based on the operative Requests for Production?  You may think this is clear to the other side; but, don’t count on it.  Likewise, don’t assume the Court shares your interpretation of the protocol.  Just settling upon an agreed-upon list of queries may not be sufficient to insure a meeting of the minds.

I suggested model language in Hit Reports:

In the spirit of cooperation and to further the goal of proportionality, we agree to run one or, within reason, more “hit reports” for you.  The purpose is to permit you to focus your subsequent requests.  We will use [name of litigation review software] and run the terms against electronically stored information in the [name database].

Please note that a “hit” may be beyond the scope of discovery, irrelevant, privileged, work product, non-responsive, a trade secret, confidential, proprietary, etc., and all rights are reserved to object to the production of any or all “hits.” This includes without limitation the right to assert that any or all production is not proportional to the needs of this case.  Further, all rights are reserved as to any and all other objections under the Maryland Rules [or Federal Rules of Civil Procedure], including the right to move for a protective order for any reason permitted under the Rules. All objections to any discovery request are incorporated herein, and none are waived by the hit report. No privilege is waived by production of a hit report because the hit report is not an offer to produce any document. Additionally, the right to mark a “hit” as confidential or highly confidential under the agreed protocol or protective order is fully reserved. Further, we reserve the right to add additional data to, or remove data from, the database and that may vary the number of hits.  All rights are reserved, including without limitation, rights under Fed.R.Evid. 502.

It may be that, in Sussman, the detailed search string minimized the risk of production of non-responsive documents; however, the scope of the order on its face appears to go beyond the scope of discovery by ordering production of non-responsive ESI.  For example, the second search string covers all Sussman custodians from 2015 to 2026.

WAS IT ERROR FOR SUSSMAN TO PRODUCE

MANY DOCUMENTS THAT DID NOT CONTAIN AGREED-UPON SEARCH TERMS?

Second,  Sussman said that it ran the agreed-upon search terms and produced approximately 800 pages.  However, “VWR states that the June 28 production cannot be what Sussman claims it to be, because many if not all of those 800 documents do not contain any of the agreed-upon search terms.”  Id. at *1 (emphasis added).

The court added that “Sussman has done little or nothing to verify the content of its production (or supplement it if necessary)….”  It ordered  that Sussman “consult with its discovery vendor as to what search terms were actually run in connection with its June 28, 2026 production” and “if necessary, run the correct searches, using the agreed-upon terms….”

The parties agreement is silent on whether the documents produced were parent-child documents.  If they were, it is not uncommon that a search term “hits” in one parent or child, but production of the parent-child email results in production of one or more documents that do not contain the search term.

I agree that explanation may not provide Sussman with a “safe harbor” here because “many if not all” of the produced documents did not contain the keywords.  And, the court noted a lack of quality control by Sussman, i.e., “has done little or nothing to verify the content.” But a little more information would be helpful.

There are some other factors that may have contributed to the court’s decision. Sussman’s production was tardy.  For example: “The Court notes that this information was due by May 14, 2026, and Sussman is now more than two months late.”  And, as to other information: “The Court notes that this information was due by March 6, 2026, and Sussman is now more than four months late.”  It may be that it was time for the curtain to fall.

This blog was initially posted on  Electronic Discovery Reference Model.