W DLC-M08-001 · Case Law Digest — eDiscovery & AI.docx — Word
FileHomeInsertDesignLayoutReferencesReviewView
DLC-M08-001 · Module 08 Deliverable
Case Law Digest — eDiscovery & AI
Compiled

The rulings that decided what "reasonable" means when the discovery is digital — and, increasingly, generated. Read in order, or jump to the section you need. Each case gets a citation, a court, the holding in plain English, and the practice implication that survives the specific facts.

Group A · Proportionality & Scope

Zubulake v. UBS Warburg LLC (I & III)
217 F.R.D. 309 (S.D.N.Y. 2003); 216 F.R.D. 280 (S.D.N.Y. 2003)
FOUNDATIONAL COST-SHIFTING
Holding: Established the modern cost-shifting framework for accessible vs. inaccessible ESI. Zubulake I introduced the seven-factor test that Zubulake III applied.
Practice implication: Sampling of inaccessible data can determine whether restoration is warranted. Preservation obligations attach when litigation is reasonably anticipated. Still the starting point for every proportionality analysis.
Rowe Entertainment, Inc. v. William Morris Agency, Inc.
205 F.R.D. 421 (S.D.N.Y. 2002)
HISTORICAL EIGHT-FACTOR
Holding: Pre-Zubulake eight-factor cost-shifting test. Superseded by Zubulake but still cited when parties argue for or against the newer framework.
Practice implication: Read for historical context; do not cite as controlling authority.
Chen-Oster v. Goldman, Sachs & Co.
285 F.R.D. 294 (S.D.N.Y. 2012)
SEARCH TERMS PROPORTIONALITY
Holding: Parties must cooperate on search-term negotiation and justify the burden their proposals impose. Court will not simply endorse a party's preferred terms.
Practice implication: Bring metrics to the meet-and-confer. Volume estimates per term, hit-count analysis, false-positive testing — proposing terms without them will not persuade.
Oracle America, Inc. v. Google, Inc. (proportionality orders)
N.D. Cal. 2015
POST-2015 FRCP PROPORTIONALITY
Holding: Applied the 2015 FRCP amendments emphasizing proportionality to the specific facts and value of the case. Rejected discovery requests as disproportionate.
Practice implication: The 2015 amendments were not cosmetic. Proportionality is now an affirmative defense to overbroad discovery.

Group B · Preservation & Sanctions

Zubulake v. UBS Warburg (IV & V)
220 F.R.D. 212 (S.D.N.Y. 2003); 229 F.R.D. 422 (S.D.N.Y. 2004)
FOUNDATIONAL LITIGATION HOLD
Holding: Preservation duty triggers when litigation is "reasonably anticipated." Counsel has affirmative duty to institute a written litigation hold and to monitor compliance.
Practice implication: Litigation hold is not optional and is not one-and-done. Track receipt, re-issue on staffing changes, and document custodian follow-up.
Pension Committee of the University of Montreal Pension Plan v. Banc of America Securities LLC
685 F. Supp. 2d 456 (S.D.N.Y. 2010)
HISTORICAL CULPABILITY STANDARD
Holding: Established that a failure to issue a written litigation hold was "gross negligence" per se — later softened by amendments to Rule 37(e) and cases like Chin v. Port Authority.
Practice implication: Still cited by opponents seeking sanctions. Be prepared to distinguish factually and note the doctrinal narrowing.
GN Netcom, Inc. v. Plantronics, Inc.
2016 WL 3792833 (D. Del. July 12, 2016)
RULE 37(e) SPOLIATION
Holding: $3 million monetary sanction plus adverse-inference instruction for intentional deletion of emails by a senior executive after litigation was anticipated.
Practice implication: Rule 37(e) after the 2015 amendments still provides meaningful teeth. Intentional deletion by senior personnel — even if hold was formally in place — is a recipe for sanctions.
Klipsch Group, Inc. v. ePRO E-Commerce Ltd.
880 F.3d 620 (2d Cir. 2018)
EVASIVE COLLECTION COST-OF-REMEDY
Holding: Second Circuit affirmed monetary sanctions equal to the cost of remedying evasive collection practices. Sanctions need not be limited to statutory awards where discovery misconduct imposed identifiable cost on the innocent party.
Practice implication: When opposing counsel's collection is evasive, quantify the remediation cost — that's your sanction ceiling.

Group C · TAR & Predictive Coding

Da Silva Moore v. Publicis Groupe SA
287 F.R.D. 182 (S.D.N.Y. 2012)
FIRST TAR APPROVAL PECK
Holding: First judicial opinion approving the use of predictive coding in eDiscovery. Judge Peck: "computer-assisted review is an acceptable way to search for relevant ESI in appropriate cases."
Practice implication: Cite when defending TAR use. The subsequent recusal motion and Rule 502(d) discussion are separately instructive.
Rio Tinto plc v. Vale S.A.
306 F.R.D. 125 (S.D.N.Y. 2015)
TAR EVOLVED PECK
Holding: Judge Peck: "the case law has developed to the point that it is now black letter law that where the producing party wants to utilize TAR for document review, courts will permit it."
Practice implication: Cite this for the "you don't need permission" position. Also read for cooperation guidance — the opinion urges transparency in TAR protocols.
Hyles v. New York City
2016 WL 4077114 (S.D.N.Y. Aug. 1, 2016)
CHOICE-OF-TOOL
Holding: Court will not force a producing party to use TAR when the producing party prefers keyword search — even if TAR would arguably yield better results. The choice of methodology belongs to the producing party.
Practice implication: Party autonomy over methodology is firmly established. Focus opposing-counsel debate on quality of output, not choice of tool.
In re Broiler Chicken Antitrust Litigation
N.D. Ill. 2018 (Order dated January 3, 2018)
TAR PROTOCOL VALIDATION
Holding: Adopted a detailed TAR validation protocol including elusion sampling parameters, stopping criteria, and transparency requirements.
Practice implication: A modern template for TAR stipulations. Adapt the validation methodology when negotiating a TAR protocol in complex multi-district litigation.

Group D · Privilege, Waiver & Clawback

Victor Stanley, Inc. v. Creative Pipe, Inc.
250 F.R.D. 251 (D. Md. 2008)
KEYWORD FAILURE WAIVER
Holding: Keyword search alone was inadequate to protect privilege where the parties failed to test terms, sample results, or apply reasonable QC. Privilege waiver where methodology was demonstrably deficient.
Practice implication: Document your keyword-testing process. Sampling and QC of privilege calls are not optional if you plan to claim reasonable care.
Mt. Hawley Insurance Co. v. Felman Production, Inc.
271 F.R.D. 125 (S.D. W. Va. 2010)
FRE 502(b) CLAWBACK
Holding: FRE 502(b) reasonableness requires an affirmative showing of the process used to prevent inadvertent disclosure. Clawback protection depends on demonstrated process, not merely intent.
Practice implication: Get a 502(d) order whenever possible. If relying on 502(b), be able to describe the QC process in detail.
FTC v. Boehringer Ingelheim Pharmaceuticals, Inc.
180 F. Supp. 3d 1 (D.D.C. 2016)
CATEGORICAL LOGS
Holding: Categorical privilege logs are acceptable in appropriate cases where document-by-document logging would be disproportionate. Court approved a categorical approach in a high-volume matter.
Practice implication: Cite when negotiating log format on productions with 10,000+ privileged docs. Draft categories carefully — they must still allow the receiving party to assess claims.

Group E · AI in Legal Practice

Mata v. Avianca, Inc.
678 F. Supp. 3d 443 (S.D.N.Y. 2023)
CHATGPT RULE 11
Holding: Attorneys sanctioned under Rule 11 for filing a brief containing fabricated case citations generated by ChatGPT. Attorneys had failed to verify the AI output before filing.
Practice implication: The sanction was not for using AI — it was for not verifying. Every citation in every AI-assisted filing gets verified. Document the verification.
Park v. Kim
91 F.4th 610 (2d Cir. 2024)
2D CIR AFFIRMANCE AI HALLUCINATION
Holding: Second Circuit affirmed sanctions against an attorney who filed a brief containing AI-fabricated citations. Rejected the argument that reliance on AI insulated the filer from Rule 11 duties.
Practice implication: AI-hallucination sanctions have appellate support. Do not assume "the AI made me do it" will fly.
Standing Orders on Generative AI (N.D. Tex., D.D.C., E.D. Pa., et al.)
Various, 2023–2025
DISCLOSURE ORDERS
Holding: An increasing number of federal judges have issued standing orders requiring attorneys to disclose AI use in filings, certify verification of AI output, or both.
Practice implication: Read the judge's standing orders before every filing. When in doubt, disclose. Some judges require an explicit certification even when not asked for.
Kohls v. Ellison & downstream AI-evidence orders
Various, 2024–2025
AI-GENERATED EVIDENCE AUTHENTICATION
Holding: Courts are beginning to grapple with authentication and admissibility of AI-generated content — including deepfakes offered as evidence and expert opinions relying on AI outputs.
Practice implication: Expect challenges when AI-generated content enters the record. Chain of custody and reproducibility documentation will be increasingly required.

Group F · Modern Communications & Ephemeral Data

Waymo LLC v. Uber Technologies, Inc.
N.D. Cal. 2018 (multiple orders)
EPHEMERAL MESSAGING WICKR
Holding: Court addressed preservation obligations for auto-deleting messaging platforms (Wickr). Uber's use of Wickr for business communications became a central issue.
Practice implication: Any client using auto-delete messaging for business communications is at preservation risk from the moment litigation is reasonably anticipated. Ask about it in custodian interviews.
Doe LS 340 v. Uber Technologies, Inc.
N.D. Cal. 2023
SLACK COLLABORATION
Holding: Court addressed preservation and production expectations for Slack and other collaboration platforms in a large-scale litigation.
Practice implication: Slack, Teams, and similar collaboration platforms are within the ordinary scope of ESI discovery. Include them in custodian interviews and production spec by default.
SEC — Off-Channel Communications Enforcement Actions
SEC settlements, 2022–2024
REGULATORY WHATSAPP SIGNAL
Holding: The SEC has imposed substantial fines (billions in aggregate) on financial services firms for failing to preserve business communications conducted on off-channel messaging platforms (WhatsApp, Signal, personal text).
Practice implication: Regulatory preservation obligations may substantially exceed litigation-hold obligations. For regulated clients, off-channel messaging is a first-line question in any assessment.

Group G · Emerging & Watch-List

Cases addressing Cross-Border Discovery & Data Privacy
Various, ongoing
GDPR SCHREMS III?
Holding: Continuing tension between US discovery obligations and EU/UK data-protection regimes. Case law is developing on the boundaries of transfer, redaction, and blocking-statute assertions.
Practice implication: On any matter with EU/UK custodians, engage local counsel early. Blocking statutes are increasingly asserted; US courts are increasingly skeptical.
Emerging AI-Related Discovery Orders
Various, 2024–2025
WATCH LIST
Holding: Multiple courts are now issuing orders addressing the discoverability of AI training data, prompts, model versions, and system logs — as well as the admissibility of AI-derived analyses. This area is developing rapidly.
Practice implication: Preserve prompts, system prompts, temperature settings, and model versions for any AI-assisted work product. Treat these as potentially discoverable.

How to use this digest

  1. For meet-and-confer preparation: pull the cases in the group corresponding to your dispute (proportionality, TAR, privilege, etc.). Read the holdings; note the courts and dates; be prepared to distinguish or apply.
  2. For internal training: assign one case per associate; hold a moot on the practice implication. Case reasoning is often more instructive than the holding.
  3. For AI-workflow risk assessment: Group E is required reading. Every attorney using AI in a workflow that touches the record should have read Mata, Park, and their local judge's standing order.
  4. For client-facing memos: quote the practice implication, not the holding. Clients want to know what to do, not what the court said.

Update cadence

Delphoria maintains this digest at 2026-Q1 as of the version noted in the header. Case law in the AI, ephemeral messaging, and cross-border areas is moving fast. Check for the latest Delphoria revision before citing in a filing.

One rule Every case in this digest lives in a factual context. The practice implication survives the specific facts; the holding may not. Read the case before citing it — always.