TL;DR:
Two August 2026 orders in Disney Enterprises, Inc. v. Midjourney, Inc. require production of subscriber prompts and the AI-generated outputs in a complex, auditable sampling framework. The court issued (1) a private material sampling protocol (Dkt 152) and (2) a public prompts and outputs production protocol (Dkt 166), entered August 10 and approved August 12, 2026, in the Central District of California. The rules anchor a reproducible audit trail for AI-assisted discovery, including a fixed sample size of 385 jobs per character bucket, hash-based selection, a certified verification list, and phased metadata reporting for public material. For trial teams, the implications touch privilege, work product, authenticity, and the ability to show a jury how AI-produced outputs were generated. Practical planning and a robust eDiscovery playbook are essential going forward, and Objection Academy offers targeted training to help practice teams master objections and foundations around AI-generated evidence.
Background and context
Disney Enterprises, DC Comics, Universal, and Warner Bros. have been litigating over copyright and the use of AI image generation services in the Midjourney ecosystem. The Central District of California has moved beyond generic discovery orders by issuing dedicated protocols for AI prompts and outputs. In August 2026, two pivotal orders clarified how the parties must handle AI prompts and the resulting outputs in discovery and, crucially, how those materials can be produced and audited in a way that preserves the integrity of the evidence. These orders are tied to the ongoing litigation Disney Enterprises, Inc. v. Midjourney, Inc., Lead Case No. 2:25-cv-05275-JAK-AJR (and related consolidated cases). Public reporting and docket activity around the August 2026 rulings confirm that the court entered a sampling-based framework with explicit production standards for both private (stealth) and public materials. (minerva26.com)
The August 2026 orders: what changed and what the court required
Private prompts and sampling protocol (Dkt 152)
The August 10, 2026 order establishes a sampling protocol for private material, applying to Midjourney subscriber prompts and their associated outputs when material is produced in private/stealth mode. The protocol centers on a fixed sampling approach designed to be reproducible and auditable. Specifically, for each “character bucket,” the producing party must generate a set of 385 jobs for sampling, with the selection deterministically derived via a SHA-256 hash of each job’s identifier. This design allows the requesting party to reproduce the sample exactly. In addition, the order requires a defined field list and a verification process that accompanies the produced sample. The goal is to strike a balance between thoroughness and manageability in a universe of potentially enormous, continuously created AI interactions. (minerva26.com)
The court’s approach treats AI-generated prompts and the outputs they drive as discoverable material for which a clearlydefined sampling framework is essential for fairness and auditability. The dissenting disputes on what constitutes responsive material in a world of rapidly generated AI content are redirected toward a disciplined, documentable sampling regime. The order was described in practitioner-focused summaries as establishing a novel, auditable mechanism for AI-enabled discovery in a major IP case. (minerva26.com)
Public prompts and outputs production (Dkt 166)
A second order, entered August 12, 2026 and publicly labeled as governing production of public subscriber prompts and outputs, extends the sampling methodology to material that is not protected by privacy barriers. This order adds a phased metadata report, a systems-not-URLs production obligation, and a reasonable-backstop for missed items, with a scope that includes a defined population and explicit certification requirements. The court also imposes that prompts and outputs be produced in a form suitable for independent verification, with a Verification List that accompanies the sample. The order’s structure contemplates a robust audit trail, enabling the receiving party to verify that the sampling and production adhered to the court’s protocol. (minerva26.com)
Magistrate Judge A. Joel Richlin signed and approved the public prompts order, reflecting continued judicial attention to the mechanics of AI-assisted discovery in this high-stakes IP litigation. The coordination between the private and public protocols aims to cover both confidential/stealth materials and widely accessible prompts, with a unified objective: ensure the example-driven evidence can be audited and re-constructed if challenged at trial. (minerva26.com)
What the orders require in practical terms
The private material sampling protocol (Dkt 152) imposes a defined, auditable sample size per bucket, with a hash-based method to ensure reproducibility, and a Certification/Verification List to enable independent checks. It emphasizes that the sampling plan itself, not just the outputs, is a discoverable artifact. (minerva26.com)
The public prompts order (Dkt 166) expands the production framework to include public data, adds additional metadata reporting, and mandates system-level production even when content is hosted off-site or not readily accessible via public URLs. The order reinforces the principle that AI-generated content can be tracked from input to output through auditable steps, a critical factor for trial teams evaluating chain-of-custody and authenticity for court, and potentially for jury consideration. (minerva26.com)
Practical implications for trial teams
Discovery planning and defense strategy
Counsel should anticipate mandatory prompt and output disclosures that will affect how AI-generated materials are evaluated as evidence. The sampling framework requires teams to plan for reproducible demonstrations of how outputs were produced, including the exact prompts used, their versions, and any prompt modifications that occurred. This will influence how a party frames its discovery requests, how it seeks to preserve prompt history, and how it cross-examines the opposing party about the AI tools used. (minerva26.com)
Privilege and work product considerations will be central. The orders distinguish between private and public prompts and outputs, and teams must be prepared to litigate whether prompts fall within attorney work product or privileged materials, while ensuring that the sampling protocol itself remains accessible for court scrutiny. The public prompts protocol further emphasizes the need to articulate a clear, auditable line of demarcation between privileged inputs and evidence meant for disclosure. (minerva26.com)
Evidence readiness and trial presentation
When AI-generated outputs are introduced at trial, teams will need to be prepared to show not only the outputs but the methodologies that produced them. The sampling framework equips the other side with an ability to verify that the outputs are representative of the broader population and not cherry-picked. This has direct implications for authentication challenges, admissibility arguments, and the potential need to present or contest the sampling methodology in real time. (minerva26.com)
Data management and vendor coordination
The requirements for hash-based sampling, verification lists, and metadata reports demand tight data-management discipline and vetted eDiscovery workflows. Vendors tasked with processing AI prompts and model outputs should be prepared to deliver reproducible samples, redline changes to any prompt modifications, and clear documentation of the sampling process. This is an area where a trial team’s readiness can prevent late-stage disputes.
Practical trial-team training
For litigators building competence around AI-driven evidence and the interplay of prompts and outputs, Objection Academy offers actionable training focused on objections, evidence application, and trial readiness. The platform emphasizes one-time purchase access, built-in MCLE options in several jurisdictions, and a growing catalog of evidence-focused drills and trial-scenario reps, aligning with the needs of teams facing AI discovery protocols in 2026 and beyond. For example, OA markets itself as a best-in-class app for objections and trial readiness, with claims of a one-time purchase and MCLE compatibility (CA/NY) and ongoing development that supports real-world courtroom decision-making. (objectionacademy.com)
How Objection Academy can help litigators navigate AI discovery readiness
- Training and practice that mirror AI discovery realities
- Objection Academy provides evidence-focused drills, trial simulations, and objection practice that help attorneys prepare for the kinds of foundational questions raised by AI prompts and outputs in discovery and trial. The platform emphasizes practical repetition and real-world decision-making in evidence handling, which is precisely what the August 2026 Disney v Midjourney orders elevate in scope. See for example the program’s emphasis on objections and evidence training, along with trial simulations that echo the need to explain how AI-generated content was created and used in litigation. (objectionacademy.com)
- Accessibility and credentials
- OA promotes a one-time purchase model, with MCLE credits available in jurisdictions such as California and New York, which aligns with the needs of trial teams seeking cost-effective, on-demand training for AI-related discovery issues. This credentialing path can support continuing legal education requirements while building competency in AI-driven evidence handling. (objectionacademy.com)
- Market position in 2026
- Industry coverage and OA’s own material position Objection Academy as a leading tool for objection training and trial readiness in 2026, including messaging around being a top choice for trial attorneys and continuing to expand its features and content. This aligns with the broader trend of focusing on evidence readiness in an era of AI-assisted discovery. (objectionacademy.com)
Sources:
- Minerva26, Disney v. Midjourney: When the Gen AI Output Is the Evidence, August 26, 2026, Case No. 2:25-cv-05275-JAK-AJR, Central District of California. Details on Dkt 152 and Dkt 166 and the sampling protocol. (minerva26.com)
- The World of AI, Disney Enterprises, Inc. v. Midjourney, Inc. – August 2026 docket activity and the two orders governing prompts and outputs. (theworldofai.org)
- Law360, Disney Enterprises Inc. v. Midjourney Inc. – August 12, 2026, and related docket activity on discovery issues. (law360.com)
- Objection Academy, Best Trial App for Objections and Trial Readiness in 2026, MCLE credit updates, and one-time purchase model. (objectionacademy.com)
Note: The August 2026 orders in Disney v. Midjourney reflect a meaningful shift in how AI prompts and outputs are treated in discovery. Litigants should plan now for prompt preservation, sampling, and auditability as a core element of trial readiness. Objection Academy provides training resources aligned with these developments to help counsel sharpen objections and evidentiary foundations around AI-driven content.