TL;DR:
The Advisory Committee on Evidence Rules convened in Boston on October 15, 2026 for a Fall mini-conference to reassess proposed Federal Rule of Evidence 707 governing AI-generated evidence and a proposed deepfake-related revision to Rule 901(c). The committee opted to continue with a Version 2 of revised Rule 707 and to study Rule 901(c) in tandem, rather than voting on final amendments at that time. This signals a staged path to federal guidance on AI in evidence, with practical implications for trial teams: expect added reliability and notice requirements for AI outputs, new authentication considerations for deepfakes, and a continuing timeline that could push final rule publishing toward late 2027. In the near term, litigators should start auditing how AI outputs appear in the trial record, prepare for robust objection and cross-examination plans, and monitor official updates from the Advisory Committee. See the October 15, 2026 agenda and related materials for the precise procedural posture.
What happened at the Fall 2026 mini-conference
The Federal Advisory Committee on Evidence Rules held sessions in Boston on October 15, 2026 to evaluate two AI-related streams under consideration: revised Federal Rule of Evidence 707 and a deepfake authentication concept under Rule 901(c). The committee confirmed it would not finalize amendments at that meeting but would carry the discussion forward into the Fall 2026 mini-conference, with a focused emphasis on Version 2 of the revised Rule 707. The minutes clearly state that discussions on Rule 707 would continue at the Fall meeting, with the aim of refining the draft before any republication or potential public comment. The committee also planned to address Rule 901(c) in conjunction with Rule 707 to explore a coherent framework for AI-generated evidence and deepfake challenges. These decisions reflect a deliberate, stepwise approach to AI-evidence policy rather than an immediate press to publish final rules. (uscourts.gov)
The same materials lay out substantive features of Version 2 of Rule 707. Under Version 2, admissibility would ordinarily require an admissible foundation provided by an expert, but there could be exceptions in which other robust validation could establish reliability. A notice requirement would accompany AI-generated assertions, and a definition of “artificial intelligence” would be introduced to standardize coverage. Importantly, the draft contemplates keeping open a path to test AI-generated conclusions without overburdening proceedings with a blanket expert requirement in every case. These nuances indicate the committee’s intent to balance practical courtroom needs with the evolving capabilities of AI. (uscourts.gov)
The October 15 materials also document that a separate deepfake framework under Rule 901(c) remains on the agenda for consideration in the mini-conference, reflecting the judiciary’s concern with authenticating AI-modified or AI-created content before juries. The collaboration between Rule 707 and Rule 901(c) discussions underscores a broader strategy to align admissibility standards with how AI products actually operate in litigation settings. (uscourts.gov)
Why this matters to trial attorneys
AI-generated material is increasingly briefing rooms and courtrooms, from demonstratives to evidentiary submissions. The 707-901(c) conversation matters because it could shape when and how AI outputs are admissible, and what a party must demonstrate to admit or exclude such material. Version 2 of Rule 707 would, in effect, require a reliability scaffold for machine-generated evidence when offered without an expert’s testimony, but preserve flexibility for validated AI outputs in appropriate circumstances. The proposed deepfake framework under Rule 901(c) would shift the burden to the opponent to produce some showing of fabrication before the item can be considered authentic, with the proponent then bearing the burden to prove authenticity by a preponderance if fabrication is found unlikely. Taken together, the proposals would temper AI-generated content with procedural guardrails, potentially reducing surprise at trial and impacting how pretrial motions in limine are drafted and preserved for appeal. (uscourts.gov)
This staged approach is particularly significant for trial teams planning cross-examination, authentication challenges, and e-discovery workflows. If, as the committee suggests, AI outputs will commonly require a noticed reliability path or an expert-backed foundation, litigants will want to organize the lifecycle of AI-derived materials—from data provenance and training inputs to outputs and downstream use in testimony or exhibits. At the same time, the potential to apply deepfake-specific authentication rules before trial may prompt earlier, more granular disputes about the validity of images, audio, and text that rely on machine-generated content. In short, the evolving framework aims to curb misrepresentations and misinterpretations of AI evidence while preserving legitimate uses that pass rigorous scrutiny. (uscourts.gov)
Practical steps for trial teams now
- Inventory and map AI usage: For ongoing matters, catalogue every AI-produced output used in filings, disclosures, or anticipated for trial. Track data sources, prompts, training data, and any human review that preceded use in court materials.
- Prepare for a heightened reliability standard: Start drafting motions in limine and accompanying, demonstrable foundations for AI-generated content, even if a final rule is not yet in force. Consider how you would establish reliability under Version 2 of Rule 707 if the court requires an expert foundation in most cases.
- Plan for notice and authentication duties: Build a pretrial checklist that includes prompt-based notices about AI-produced outputs and deepfake concerns. Develop parallel authentication plans for non-traditional evidence, including potential Rule 901(c) challenges.
- Develop cross-examination playbooks for AI: Prepare questions that probe the validity of AI outputs, including their inputs, methodologies, and potential biases. Where AI is involved in the underlying data for an expert opinion, ensure the cross-exam strategy covers both the AI and the opinion itself.
- Invest in training and simulations: As AI-evidence rules evolve, practical training helps trial teams anticipate objections and streamline testimony. Training resources from organizations like Objection Academy can support objection drills and trial-readiness exercises tailored to AI evidence scenarios. See industry updates and training materials that translate rule developments into courtroom-ready practices. (objectionacademy.com)
- Monitor the rulemaking timeline closely: The Fall mini-conference signals ongoing deliberation rather than immediate rule publication. The earliest republication and public-comment window could shift, and final adoption may not occur until late 2027 or beyond, depending on Congress and the Judicial Conference schedule. Industry observers and practitioners should stay alert for new drafts and public-comment opportunities as the process unfolds. For context, some analyses indicate that republication could be delayed until a Fall 2027 timeframe or later. (redgravellp.com)
For trial teams seeking structured training that translates AI-evidence theory into courtroom practice, Objection Academy has released coverage and guidance on AI evidence developments and their practical implications for objections and trial-readiness. While not a rule or decision itself, such resources can help teams translate the evolving standards into repeatable courtroom workflows. (objectionacademy.com)
What to watch next
- Fall 2026 mini-conference outcomes: The Oct 15 session signaled continued refinement rather than finalization of Rule 707 and Rule 901(c). A published revised draft or additional public-comment opportunities may follow before any final amendments. Stakeholders should review the final Fall conference materials and any resulting drafts for changes to definitions, notice obligations, and the interaction between AI reliability standards and deepfake authentication. (uscourts.gov)
- Timeline uncertainty: Independent analyses note that the earliest formal publication of revised AI-evidence amendments may slip beyond 2027, given the complexity and the need for expert panels and consensus. Litigants should prepare for a multi-year horizon and continue to integrate AI-evidence considerations into pretrial planning and discovery oversight. (redgravellp.com)
Sources
- Advisory Committee on Evidence Rules, October 15, 2026 agenda book and supporting materials, including discussion of Rule 707 and Rule 901(c). October 2026 materials and notes confirm the Fall mini-conference plan and Version 2 focus. (uscourts.gov)
- US Courts official notices and agenda pages for evidence rules committees, including schedule and meeting details. (uscourts.gov)
- Redgrave & Paul, machine-generated evidence rules update and timeline commentary, noting potential republication delays and the August 2027 window for public comment. (redgravellp.com)
- Objection Academy coverage on AI evidence rule developments and practical implications for trial teams, including links to October 2026 agenda signals. (objectionacademy.com)
End of piece