TL;DR:
In Crowder et al v LinkedIn Corporation, the Northern District of California issued a July 31, 2026 discovery order that curtails AI-assisted hyperlink discovery by limiting LinkedIn to producing up to 100 hyperlinked documents identified by plaintiffs. The ruling clarifies that AI-enabled discovery remains governed by traditional Rule 26 standards, with a careful, proportional approach to data requests. The decision also preserves a pathway for targeted fact-finding through a controlled discovery schedule, including a potential Rule 30(b)(6) deposition of LinkedIn. For trial teams, the order emphasizes disciplined, documents-first discovery and cautions against broad, burdensome AI disclosures. This development directly affects anticipatory discovery planning in antitrust and tech-adjacent matters and signals how AI-enabled workflows will be integrated into standard TAR practices going forward. Trial teams should align ESI protocols, privilege handling, and cross-examination plans to reflect a measured approach to AI tooling and hyperlink data. Objection Academy’s training resources on AI in discovery and objection handling provide practical playbooks for implementing these standards in the courtroom.
Overview of the dispute and the action
The Crowder case, Crowder et al v. LinkedIn Corporation, No. 4:22-cv-00237-HSG (LB), centers on LinkedIn’s alleged anticompetitive practices and data practices. A dispute arose over whether LinkedIn must search for and produce documents referenced by hyperlinks embedded in emails produced in custodial data. The court’s July 31, 2026 order, signed by Magistrate Judge Laurel Beeler, resolves that dispute by adopting a compromise: LinkedIn must investigate and produce up to 100 hyperlinked documents identified by the plaintiffs, to the extent available and not already produced or privileged. If a requested item is unavailable, LinkedIn must confirm its unavailability in writing. This ruling follows extensive meet-and-confer sessions and builds on a July 1, 2026 order addressing Relativity aiR usage for document review, situating AI-enabled review squarely within traditional TAR frameworks. The July 31 order also notes that the parties had engaged in substantial discovery discussions throughout May and June 2026 and that the court can decide the dispute without oral argument. (docs.justia.com)
A parallel thread in the same docket confirms LinkedIn’s use of Relativity aiR for final responsiveness determinations, a development described as a first-of-its-kind federal ruling validating AI-assisted review under familiar TAR standards. The court’s approach treats GenAI-driven review as a technology-assisted process rather than as a wholly new regime, reinforcing the applicability of reasonableness, proportionality, and burden-shifting under Rule 26. Several reputable law firms and industry commentators summarized Schulte as establishing a practical framework for AI-enabled discovery that trial teams can implement going forward. (dlapiper.com)
What the order does and does not do
- Cap on hyperlinked discovery: LinkedIn is required to search for and produce up to 100 hyperlinked documents identified by plaintiffs, provided those documents are available and not privileged. This cap reflects the court’s attempt to balance relevance with the burdens of hyperlinked data across multiple hyperlinked items in emails. The order explicitly states that if a requested document has already been produced or is privileged, LinkedIn should identify it by Bates number or privilege log entry, and if not available, confirm in writing. (docs.justia.com)
- Consistency with prior AI discovery rulings: The court’s July 31 order cites the July 1, 2026 decision regarding Relativity aiR as a basis for treating AI-assisted review as a TAR-like process subject to traditional discovery rules rather than creating a separate AI-discovery regime. This framing preserves a familiar evidentiary calculus for litigants while incorporating GenAI into the workflow. (docs.justia.com)
- Discovery process craftsmanship: The order emphasizes proportionality, the need to meet and confer, and a targeted approach to AI-enabled requests. It references the parties’ extensive meet-and-confer history and allows for a measured path forward, including the potential for additional cooperation on the scope of AI-assisted review as the case progresses. The court also signals openness to refining discovery parameters through further negotiation, rather than unilateral mandates. (docs.justia.com)
- Depositions and topics: The order confirms that plaintiffs may take up to fifteen depositions of LinkedIn witnesses, including one Rule 30(b)(6) deposition, with a framework to coordinate overlapping Rule 30(b)(1) and Rule 30(b)(6) witnesses. This preserves meaningful fact-finding opportunities while avoiding unbounded discovery. (docs.justia.com)
Practical implications for trial teams
- AI workflows stay within established TAR paradigms: The Crowder order reinforces that generative AI tools used in document review, such as Relativity aiR, should be evaluated through the same lenses of recall and precision, sampling, and validation that have long governed TAR workflows. This reduces the risk of creating a separate, opaque AI-discovery regime and helps trial teams normalize AI-assisted review as part of standard discovery practice. (dlapiper.com)
- Targeted production over broad sweeps: The 100-hyperlink cap pushes parties toward focused, relevant production rather than sweeping, indiscriminate extractions. Counsel should structure hyperlinked discovery requests with clear identifiers, lookback windows, and defensible search strategies to maximize relevance while minimizing burden. Judges are signaling that proportionality matters, especially in data-rich tech disputes. (docs.justia.com)
- Discovery planning and ESI protocol: As courts embrace AI tools within traditional discovery, litigants should codify AI usage in a robust ESI protocol, including data preservation, search term governance, sampling plans, and auditability. This helps ensure that AI-driven results can be defended on appeal and in cross-examination, aligning with the TAR-centric approach the court endorsed. (dlapiper.com)
- Cross-examination and evidence-readiness: The cap and the TAR framing provide a clear basis for cross-examining opposing counsel about AI-assisted decisions, including the methodology used by Relativity aiR, the reproducibility of results, and the adequacy of validation sampling. The Crowder ruling underscores the importance of developing a concise cross-exam plan that tests AI-driven outcomes without demanding excessive disclosure of proprietary algorithms. (dlapiper.com)
- Practical steps for defense and plaintiffs:
- Build a precise discovery plan specifying the population of custodians, search terms, and the scope of hyperlinked data to be reviewed.
- Prepare a request for calendar-efficient timelines, noting the 100-document cap as a predictable throttle on AI-driven review.
- Schedule a targeted Rule 30(b)(6) deposition to address topics such as LinkedIn’s use of AI review platforms, governance controls, and QA processes.
- Maintain a privilege and withholding log to accompany any produced or omitted hyperlinked items.
These steps reflect the court’s emphasis on controlled, transparent, and proportionate discovery governing AI-enabled workflows. (docs.justia.com)
How Objection Academy helps with timely AI discovery developments
Timely developments like the Crowder decision demand practical training for trial teams on AI-assisted discovery and objection handling. Objection Academy has published contemporaneous analysis and practical drills on AI-assisted discovery, including how courts treat AI review within TAR frameworks and how to structure effective cross-examinations around AI-derived evidence. Training materials emphasize precision, proportionate discovery, and courtroom-ready objections tailored to technology-driven data. The platform reports on recent developments and provides scenario-based practice to prepare litigators for rapid changes in eDiscovery practice. (objectionacademy.com)
Next steps for trial teams
- Incorporate AI discovery into the ESI plan: Draft a data plan that accounts for AI review methodology, sampling, validation, and the ability to supplement the administrative record if needed.
- Prepare for controlled hyperlinked production: Identify target hyperlink sources, designate custodians, and map a plan for producing up to 100 hyperlinked documents with clear Bates and privilege-log references.
- Align briefs and voir dire with AI discovery strategy: Use the 100-document cap and TAR-like standards to frame objections, questions about AI-driven processes, and cross-examination lines about data provenance.
- Leverage practical tools and training: Engage with Objection Academy and other reputable resources to rehearse AI-driven discovery scenarios and objection handling in a way that translates to courtroom performance.
Sources:
- Crowder et al v LinkedIn Corporation, No. 4:22-cv-00237-HSG (LB), Discovery Order addressing 220 Discovery Letter Brief (July 31, 2026). Justia docket: 225. (Primary order) (docs.justia.com)
- Crowder et al v LinkedIn Corporation, 4:22-cv-00237-HSG (LB), Order (July 31, 2026) addressing hyperlinked discovery. Justia docket: 233. (law.justia.com)
- Schulte v LinkedIn Corp., No. 22-cv-00237-HSG (LB), Discovery Order (June 30, 2026) and related summaries describing Relativity aiR for final responsiveness determinations. See Baker McKenzie summary and DLA Piper coverage. (dlapiper.com)
- Relativity press release and industry analysis of Relativity aiR in 2026, noting AI-enabled reviews are now standard in RelativityOne. (relativity.com)
- Objection Academy: AI discovery and related developments, including articles on Schulte and Crowder coverage. (objectionacademy.com)