Back to Articles

AI Evidence Rules Update: Federal Advisory Committee Reconsiders Rule 707 and Deepfake Authentication Rule 901(c)

TL;DR:

A major federal evidence-law update is unfolding around AI and deepfakes. The Advisory Committee on Evidence Rules convening on October 15, 2026 in Boston will discuss a revised approach to machine-generated evidence under proposed Rule 707 and a two-step deepfake authentication framework under proposed Rule 901(c). The committee previously withdrew Rule 707 for further study and has not finalized any amendments, but a fall mini-conference is expected to yield new draft proposals. Separately, federal courts are already testing AI discovery protocols, with Disney Enterprises v. Midjourney in the Central District of California issuing August 2026 orders that govern production of AI prompts and outputs. Collectively, these developments signal stricter scrutiny of AI evidence, new discovery obligations, and sharpened motivational to build robust trial-readiness playbooks now.

What the October 15 2026 agenda signals for trial lawyers

The Advisory Committee on Evidence Rules has published an agenda for its October 15, 2026 meeting that confirms a dual focus on AI in evidence practice. First, the committee has continued its work on a proposed new Rule 707 aimed at machine-generated evidence offered without an accompanying expert. The agenda notes that the committee originally approved a public-comment version of Rule 707 in Spring 2025, then withdrew that proposal for further study, and has since considered a modified version for continued examination. The panel discussion on AI and evidence is explicitly part of the day’s program. This is not a final rule yet, but it represents a formal, high-level invitation to practitioners to weigh in as the committee tests how courts should handle AI-derived evidence in the absence of expert testimony. (Source: Advisory Committee on Evidence Rules, October 15, 2026 agenda, pages describing Rule 707 and the AI panel.) (uscourts.gov)

Second, the agenda confirms ongoing deliberations on a proposed Rule 901(c) that would govern deepfake evidence. The draft rule envisions a two-step process: the opponent must present enough evidence to justify a court inquiry into whether an item is a deepfake; if a court finds that step warranted, the proponent would then bear a heightened burden to prove that the item is authentic by a preponderance of the evidence. The committee emphasizes that broad, blanket claims of “deepfake” are not enough to trigger the inquiry; a showing of specific concerns about manipulation is required, and only then would the court demand a rigorous authenticity showing. The document also notes that 901(c) would apply to items offered under Rule 901 or 902. This two-step approach, if adopted, would raise the bar for admitting AI-altered or AI-produced content and require counsel to be prepared with robust authentication and data-verification strategies. (Source: USC agenda book, October 15, 2026; details on 901(c).) (uscourts.gov)

Taken together, these items reveal an inflection point: federal judges and lawmakers are moving toward structural safeguards for AI-generated or AI-derived evidentiary materials, with a clear preference for explicit, practice-ready standards rather than ad hoc treatment at trial.

Disney v. Midjourney: a practical precursor for AI discovery in court

Aug 2026 brought concrete discovery protocols in AI-copyright litigation. In Disney Enterprises, Inc. v. Midjourney, Inc., the Central District of California issued two August 2026 orders governing discovery of subscriber prompts and outputs from a generative AI platform, establishing a sampling framework for private prompts and a separate protocol for public prompts, with a scheduling order to address remaining training-data issues. The orders translate into an enforceable blueprint for prompt-history preservation, prompt-logs, and the kind of data governance that trial teams will increasingly be asked to produce in AI-related disputes. While the Disney case centers on copyright and training data, the practical implications for trial teams are broad: courts may require disclosure of prompts, system prompts, and related metadata, and parties should prepare to defend or challenge the provenance and training-data relevance of AI-generated outputs. (Sources: Disney v Midjourney docket coverage and contemporaneous commentary noting August 2026 discovery orders; see also coverage tying the orders to a broader AI-discovery framework.) (dockets.justia.com)

This real-world example demonstrates a shift from piecemeal objections to a formal, court-ordered framework for AI materials. It also underscores the need for litigants to synchronize discovery plans with technical realities of AI systems, including prompt logs, model prompts, and training-data implications.

Practical implications for working litigators

  • Expect heightened authentication scrutiny. If Rule 901(c) progresses, courts may require a threshold showing that AI-generated or AI-enhanced content could be a deepfake before the proponent must establish authenticity by a heavier standard. This changes when and how counsel can offer AI-based visuals, audio, or text, and it elevates the importance of meticulous chain-of-custody and source disclosures. (Source: USC Evidence Rules Agenda, Rule 901(c) discussion.) (uscourts.gov)

  • Prepare for AI-focused discovery protocols. Disney v. Midjourney demonstrates that courts may order structured production of prompts and outputs, including sampling of private prompts and a separate treatment for public prompts. For trial teams, this means building a prompt-logs inventory, mapping custodians to AI systems, and incorporating these protocols into protective orders and Rule 26 disclosures. It also suggests that training data and model provenance may enter the record in a way that affects damages, infringement, or credibility analyses. (Sources: Disney v Midjourney orders; commentary describing August 2026 protocols.) (objectionacademy.com)

  • Integrate AI-evidence readiness into trial-readiness programs. Practical objection training and trial simulations should incorporate AI-evidentiary scenarios, including objections to AI-generated outputs, authentication hurdles, and cross-examination strategies for experts relying on AI tools. Objection Academy already positions itself as a useful resource for objection drills and trial-readiness exercises, which aligns with the evolving landscape described by the committee’s October 2026 agenda and real-world Disney/Midjourney developments. (Sources: Objection Academy coverage; Rule 707/901(c) discussions in the USC agenda.) (objectionacademy.com)

  • Protect work product and privilege in AI contexts. The Disney orders distinguish between prompts used for litigation strategy and those tied to fact discovery, suggesting that counsel should design privilege review workflows that separate material that is protectable from material that is discoverable. This nuance is critical as AI tools increasingly blur the line between attorney work product and client data. (Source: Disney v Midjourney orders and related commentary.) (objectionacademy.com)

  • Actively shape protective orders around AI discovery. Given the Disney precedent and the anticipated Rule 707/901(c) developments, counsel should consider proposing protective orders that align with the Disney/Midjourney protocols as a baseline, then tailor the terms to the case-specific custodians and systems. This reduces later disputes about scope and ensures a coherent data-management plan from the outset. (Source: Objection Academy synthesis and Disney/Midjourney context.) (objectionacademy.com)

Next steps for trial teams

  • Monitor the October 15, 2026 mini-conference outcomes. The committee’s deliberations could yield revised drafts that move toward public comment or standing committee approval. Practitioners should track the final language and readiness of Rule 707 and Rule 901(c) to assess how to adjust discovery and evidentiary strategies in federal cases. (Source: USCAGENDA, October 15 2026.) (uscourts.gov)

  • Initiate prompt-log preservation and AI-discovery planning now. Start with a practical inventory of likely AI systems involved in the litigation, document the types of prompts used, establish retention timelines, and align with protective-order language mirroring the Disney approach to private versus public prompts. This proactive approach reduces risk of disputes and supports faster motion practice if an authenticity challenge arises. (Sources: Disney discovery orders; USC evidence rules discussions.) (objectionacademy.com)

  • Integrate AI-evidence readiness into CLE and internal training. As these rules evolve, keeping trial teams current with objection strategies, authentication standards, and case-management implications will be essential. Training resources such as Objection Academy can help teams prepare to handle AI-based objections and demonstrations in court. (Sources: Objection Academy materials; USC agenda content.) (objectionacademy.com)

  • Prepare to adjust trial-facing strategy for future AI evidence. Until Rule 707 and 901(c) are finalized, practice planning should incorporate flexible scripts for objections to AI outputs, priority on source disclosures, and cross-examination frameworks for AI experts that address model provenance and training data. The trajectory is toward more disciplined, auditable AI-evidence handling in federal courtrooms.

Sources

This timely convergence of official rulemaking and real-world discovery orders makes AI evidence a central litigation issue for trial teams in 2026 and beyond. By aligning discovery plans, authentication readiness, and trial-readiness training with the evolving framework, litigators can position objections and evidentiary strategies to respond effectively to AI-driven challenges in the courtroom.

More Legal InsightsExplore MCLE Eligibility