TL;DR: In a July 14, 2026 order, Judge Katherine Polk Failla of the Southern District of New York granted a motion to seal production material comprising data repositories in the Dow Jones & Company, Inc. and NYP Holdings, Inc. v. Perplexity AI, Inc. case, limiting public access to critical AI discovery data. The order also directs sealing of related docket entries and reflects the ongoing contention over the scope and burden of AI discovery in live litigations. This development affects how trial teams plan discovery, present evidence, and handle confidential AI data in courtroom settings, including cross examination and exhibits. It also signals heightened protective measures around large AI data productions, with substantial implications for e discovery budgeting and workflow. (docs.justia.com)
Background
Dow Jones & Company, Inc. and NYP Holdings, Inc. (News Corp. affiliates) sued Perplexity AI, Inc. alleging copyright and related claims tied to Perplexity’s AI answer engine and its handling of Dow Jones and New York Post content. The case has evolved into a sprawling discovery dispute involving massive data repositories, including the Retrieval-Augmented Generation data and the Log Database, requiring substantial engineering effort to extract and review. The dispute has drawn scrutiny from practitioners for how large-scale AI data should be produced and safeguarded in litigation. For context, counsel representing Dow Jones and NYP Holdings pressed for seven additional months of data repository snapshots, a request that triggered a detailed cost-burden analysis and a judicial ruling on proportionality and secure handling. (docs.justia.com)
What happened in July 2026
- July 8 and July 10 endorsements by Judge Failla preceded the July 14 sealing order and framed the ongoing discussion about data production and protective measures in this matter. The court’s later order explicitly ties to those endorsements and directs that the sealing actions be implemented. This sequence illustrates how the court gradually narrows public access while allowing important discovery to proceed under protective terms. (docs.justia.com)
- July 14, 2026, the court entered an order terminating a letter motion to compel and granting a sealing motion related to the production of data repositories. The order also indicates that docket entries 166 and 167 should be kept under seal, underscoring the court’s intent to maintain confidentiality around sensitive AI data assets and related materials. (docs.justia.com)
- The underlying dispute involves the practical burden of producing AI data at scale. In April 2026, the court noted the substantial engineering resources and costs already expended in producing large data sets, including over 100 terabytes of Log Database data, with transfers taking many hours per day and storage costs running into tens of thousands of dollars per month. The court’s later sealing decision interacts with this context by restricting access to the most sensitive discovery materials. (docs.justia.com)
- Coverage from industry outlets reinforced the broader significance of the ruling. Bloomberg Law reported on the related development that Perplexity was compelled to search the founders’ personal emails in the IP dispute, illustrating the intensity of discovery in AI-related IP cases at the time. This pairing of developments highlights the evolving landscape of AI discovery where court orders balance transparency with confidentiality for high-stakes data. (news.bloomberglaw.com)
Practical implications for trial teams
- Heightened protective orders and data secrecy: The sealing of AI data repositories confirms that researchers and litigants must anticipate protective orders that cover not only documents but also the raw data and data structures underpinning AI systems. Trial teams should plan to present or rely on sealed exhibits, or publicly redacted representations, when necessary to support arguments or cross-examinations. This increases the need for careful exhibit planning and clear in-court narratives around what is sealed and what remains admissible. (docs.justia.com)
- Discovery budgeting and logistics: The April 2026 decision outlining the cost of generating seven additional months of log snapshots—roughly $15,000 per month for storage, with transfer times of hours per day and multi-month extraction timelines—illustrates the proportionality calculus courts apply in AI discovery. Litigation budgets should account for the substantial engineering and data-management resources required to produce AI data, especially when such data involves proprietary system internals. The sealing order, which constrains public access, does not eliminate these costs but may influence how they are allocated and justified to the court. (docs.justia.com)
- Impact on cross examination and admissibility: When critical discovery materials are sealed, trial teams must adapt cross-examination strategies to avoid reliance on publicly viewable data while still offering strong, credible avenues to challenge or defend the underlying claims. This may include focusing on authenticated, court-accessible summaries or redacted data sets, along with testimony that can be vetted for reliability without exposing sensitive system details. Objection handling and trial-readiness skills remain essential here, and practical training on AI-driven evidence, authentication, and seal management can help trial teams maintain courtroom effectiveness. Objection Academy can support teams in building objection strategies around sealed AI data and preserving confidentiality while preserving trial impact.
- Broader implications for AI litigation strategy: Sealing orders in AI discovery reflect a growing trend toward balancing transparency with legitimate business confidentiality in AI disputes. Attorneys should consider how to structure protective orders at the outset of AI-related litigation to minimize later disputes over access, scope, and proportionality, and to enable smoother trial presentation if sealed data becomes a central issue.
Practical steps for practicing attorneys
- Anticipate protective orders for AI When drafting or negotiating protective orders, specify what constitutes confidential AI data, how it can be accessed by teams, and what public filing formats are permissible. Include a plan for redacted or in-camera review of sealed materials for trial use.
- Budget for AI data workflows: Build budgets that reflect the cost of large-scale data extractions, data storage, data curation, and secure handling. Coordinate with eDiscovery vendors early to scope the data footprint and the proposed protective measures.
- Plan cross examination around sealed Develop a strategy that relies on admissible, court-facing materials and witness testimony that can be credibly challenged or supported without exposing sensitive data. Practice objections and foundation laying for AI-generated data, including authentication and hearsay concerns, using realistic, courtroom-ready drills.
- Leverage training resources for trial teams: Objection Academy offers targeted practice for objections and trial readiness in the context of complex evidence, including AI-driven data and sealed materials. By simulating objections to AI data handling and to sealed exhibits, trial teams can improve courtroom performance without compromising confidential information.
- Build a documented data-handling narrative: Prepare clear, defensible explanations of why certain AI data is sealed, how it was produced, and why redacted or summarized forms are appropriate in court. This helps lessen the risk of unnecessary public disclosures and supports strategic decision-making during trial.
Objection Academy in the current AI discovery landscape
Objection Academy remains a relevant tool for trial teams navigating the unique challenges posed by AI data and sealed discovery materials. Its objection drills, trial simulations, and evidence training can help teams anticipate objections to AI-generated content, manage the appearance and use of sealed data in court, and cultivate practical courtroom readiness. In a climate where AI cases increasingly involve extensive technical data and protective orders, practical, focused preparation translates into stronger advocacy on the record.
Conclusion
The July 14, 2026 sealing order in Dow Jones & Company, Inc. et al v. Perplexity AI, Inc. marks a notable milestone in AI discovery. By sealing production data repositories and related materials, the Southern District of New York acknowledges the sensitive, large-scale nature of AI data while preserving the integrity of the judicial process. For trial teams, this means recalibrating discovery planning, budgeting for data workflows, and sharpening cross-examination and objections around sealed AI evidence. As AI litigation continues to evolve, the ability to present credible, well-supported arguments using redacted or court-approved data will be a decisive factor in courtroom effectiveness. Practicing attorneys should monitor developments in this and related cases, and consider integrating structured AI-data training into ongoing trial-readiness programs.
Sources:
- Dow Jones & Company, Inc. v. Perplexity AI, Inc., No. 24-cv-07984 (S.D.N.Y.). Order granting sealing of data repositories dated July 14, 2026. Justia docket entries summarizing the endorsement sequence and seal order. (docs.justia.com)
- Dow Jones & Company, Inc. v. Perplexity AI, Inc., No. 24-cv-07984 (S.D.N.Y.). Motion history including anticipated seven additional months of log data production and the related burden analysis. April 6, 2026 memo endorsement. (docs.justia.com)
- Coverage: Bloomberg Law reporting on the personal emails discovery in the same IP dispute, July 10, 2026. (news.bloomberglaw.com)
- Coverage: MLex reporting on the access to founders’ personal emails in the same matter, July 8, 2026. (mlex.com)
- Additional contemporaneous reporting and docket updates available through Justia docket entries and related sources. (cases.justia.com)