TL;DR:
A June 26 2026 order in Encyclopaedia Britannica, Inc. v. Perplexity AI, Inc., SDNY No. 25 Civ. 7546 (JLR), directed a structured update to discovery and data handling around AI-enabled review. The court required transcripts of a discovery conference, set out a data-review timetable tied to RAG tooling and user logs, directed finalized ESI search terms and custodians, and extended deadlines for admissions and interrogatories while scheduling a status conference. The ruling signals a front-line shift in how courts govern AI-driven discovery, including the need to preserve prompts, model outputs, and source code through active case management. Trial teams should integrate explicit AI governance into discovery plans, prepare to obtain and safeguard AI provenance data, and align objections and cross-examinations with the evolving treatment of AI-assisted work product. Objection Academy can help trial teams practice objections and trial-ready responses to AI-related evidence and discovery disputes in this new environment.
What happened in Encyclopaedia Britannica, Inc. v. Perplexity AI, Inc.
On June 26, 2026, United States Magistrate Judge Sarah L. Cave issued an order in Encyclopaedia Britannica, Inc. v. Perplexity AI, Inc., SDNY Case No. 25 Civ. 7546 (JLR), following a discovery conference about data produced and processed with AI tools. The order lays out concrete steps and dates to manage AI-driven discovery in a high-stakes technology and information context. Key elements include:
- Transcript procurement: By June 30, 2026, the parties were to order a copy of the conference transcript.
- Data Review plan: The Dow Jones Snapshot Retrieval-Augmented-Generation (RAG) data and related user logs were to be produced on a set schedule (for example, anticipated production around July 7, 2026), with hosting options and any cost shifting to be discussed.
- ESI protocol: The parties were ordered to meet and confer to finalize search terms and custodians, with a view toward substantial completion.
- Source code and related materials: The court directed continued meet-and-confer discussions regarding potential source-code requests.
- Scheduling and thresholds: The deadline for requests for admission and interrogatories was extended to August 18, 2026.
- Status conference: A telephone conference to discuss discovery progress was scheduled for July 23, 2026.
- Context: The order follows a broader wave of attention to how AI-powered tools and data sources are used in litigation, including how to handle data provenance, model outputs, and the logs that underlie AI-generated results.
These directives appear in Justia’s docket entry for the order (D. Conn./SDNY proceedings in Encyclopaedia Britannica, Inc. v. Perplexity AI, Inc.), which reflects the court’s effort to bring precision and accountability to AI-driven discovery. The docket entry shows the explicit sources and dates referenced in the order, including the June 26, 2026 date and the June 30 transcript-order instruction, with subsequent scheduling of a July 23 conference. (docs.justia.com)
For broader context on the decision framework and protective-order practice shaping these AI-discovery issues, practitioner coverage and commentary highlight the trend toward explicit governance around AI tools in discovery and the growing demand for transparency around AI usage in expert work and data processing. See analyses and practitioner guides discussing protective orders, AI tool restrictions, and the importance of work-product arguments when AI-generated materials are at issue. (kirkland.com)
Why this matters for trial teams
Practical shift in discovery management: This order shows federal courts actively coordinating AI-related discovery early in the case. Rather than waiting for disputes to erupt, judges are issuing structured timetables for data review, log preservation, and source-code consideration. This reduces later fights over what AI-generated materials are permissible or discoverable and makes it easier for teams to plan, budget, and allocate resources.
AI provenance and transparency: Requiring data such as RAG-derived data sets and user logs to be produced or reviewed signals that courts want visibility into the mechanics of AI-assisted analysis used to generate or filter evidence. For trial teams, this elevates the importance of maintaining traceable workflows for AI-assisted review and for cross-examination of AI-derived conclusions.
ESI and source-code discipline: The order pushes the parties to define search terms and custodians up front, and to address potential source-code production. This reinforces the need to include precise AI governance language in protective orders and discovery plans, including whether and how AI tools may be used on confidential information and what logs may be produced.
Scheduling discipline and trial-readiness: An extended admissions/interrogatories deadline and a scheduled status conference create predictable milestones that help trial teams align discovery with other trial phases. It also provides a framework for using AI-aware discovery milestones in other cases.
Training and objection-readiness implications: The rise of AI-guided discovery means litigators will increasingly confront objections and evidentiary challenges related to AI outputs, prompts, and model behavior. The procedural clarity in this order underscores the need for robust objection-handling strategies and the ability to articulate the admissibility and reliability of AI-derived materials during trial.
Practical steps for practicing litigators
Update discovery plans to cover AI workflows:
Define which AI tools may be used, by whom, and for what purposes.
Require preservation and disclosure of AI prompts, model prompts, and any intermediate inputs and outputs that fed into produced materials.
Specify data sources to be reviewed (ESI, logs, data snapshots, source-code repositories) and the format of production.
Build AI governance into protective orders:
Prohibit or tightly regulate training on confidential materials using AI tools.
Require disclosure of the AI tools used, including versioning and provider details, and compel maintenance of an audit trail for AI-assisted work products.
Clarify whether prompts, prompts’ logs, and model outputs qualify as attorney work product or confidential information, and how they will be treated at trial.
Prepare for AI-related witness testimony and cross-examination:
Develop lines of questioning around methodology and data provenance of AI-assisted analyses.
Anticipate objections asserting or limiting reliance on AI outputs, and craft evidence-based responses to defend the admissibility and reliability of AI-generated conclusions.
Leverage training resources for objections and trial-readiness:
Objection Academy can help trial teams simulate objections to AI-derived evidence, practice preserving privilege when AI tools are involved, and rehearse cross-examination strategies for AI-assisted analyses. Training in this area supports the ability to address issues such as authenticity, chain of custody, and reliability of AI-generated results in real time at trial.
Coordinate with technology and e-discovery vendors early:
Align vendor capabilities with the court’s expectations for data hosting, log capture, and prompt-trailability.
Ensure that the team has access to logs, data lineage, and any necessary source-code artifacts in a controlled, reviewable manner.
The broader takeaway for 2026 and beyond
The Encyclopaedia Britannica v. Perplexity AI order illustrates a clear trend: courts are increasingly treating AI-driven discovery as a first-order governance issue. The explicit management of data review artifacts, search terms, custodians, and potential source-code material signals that AI-enabled litigation will demand deeper procedural discipline and more granular preparation than in the past. For practitioners, the implication is straightforward—proactively address AI in your discovery plans, protective orders, and trial strategy, and build your team’s capability to navigate AI-related objections and evidentiary challenges at trial.
As a practical companion to this evolving landscape, trial teams can rely on structured training and drills that mirror the courtroom realities of AI-enabled evidence. Objection Academy provides targeted practice in objection handling, cross-examination, and evidence readiness in contexts where AI tools intersect with litigation, helping sophisticated litigators stay ahead of the curve in AI-informed disputes.
Sources:
- Encyclopaedia Britannica, Inc. v. Perplexity AI, Inc., No. 25 Civ. 7546 (JLR) (S.D.N.Y. June 26, 2026) Order (Justia docket entry 78). (docs.justia.com)
- Encyclopaedia Britannica, Inc. v. Perplexity AI, Inc.: discovery updates and conference scheduling (Justia docket text and order). (docs.justia.com)
- Conservation Law Foundation, Inc. v. Shell Oil Co., No. 3:21-cv-00933 (D. Conn. 2026): magistrate order compelling AI prompts production (May 18, 2026). Dechert Re:Torts coverage confirms the discovery development and its significance. (dechert.com)
- Additional context on AI protective orders and discovery management (Morgan v. V2X, Inc.; Jeffries decisions) and practitioner primers on AI in discovery. (law.justia.com)