Machine Evidence Faces New Reliability Hurdle Under Proposed Federal Rule 707

Machine output keeps showing up in modern cases with a familiar litigation move: label the result “just software,” sponsor the exhibit through a custodian-style witness, and treat reliability as a closing-argument issue instead of a gatekeeping issue. Proposed Federal Rule of Evidence 707 targets that move. The draft would require a Rule 702-style reliability showing when machine-generated evidence is offered without an expert witness and would be subject to Rule 702 if a witness testified to the same conclusion. Public comment remains open through Feb. 16, 2026, which makes January 2026 a practical moment for litigators because the committee record already reads like tomorrow’s motion briefs.

Black Box Meets Daylight

Proposed Rule 707 is short on purpose and long on consequences. The draft provides that when machine-generated evidence is offered without an expert witness and would be subject to Rule 702 if testified to by a witness, the court may admit the evidence only if the evidence satisfies the requirements of Rule 702(a)-(d). The draft also states that the rule does not apply to the output of “simple scientific instruments.”

Committee materials explain why the proposed rule points directly to Rule 702 rather than creating a separate reliability test. The Standing Committee agenda book describes the Advisory Committee’s view that reliability concerns around machine-generated output can resemble the reliability concerns courts already manage for expert testimony, and that importing familiar Rule 702 standards may reduce reinvention while keeping the focus on reliability.

Rule 702 supplies the familiar structure. Under Rule 702, the proponent must satisfy the rule’s requirements for helpfulness, sufficient facts or data, reliable principles and methods, and reliable application of those principles and methods.

Proposed Rule 707 matters because it tries to prevent a procedural sidestep. Committee materials state that parties should not be able to evade Rule 702’s reliability requirements by offering machine output directly, or through a witness who used a system but knows little about its reliability, when the output would be treated as expert testimony if the conclusion came from a human expert.

Framing Decides the Fight

Rule 707 litigation will start with characterization. The trigger turns on whether the machine output, if stated by a witness, would be subject to Rule 702. Committee materials frame the target as output that functions like expert inference rather than routine recording. That framing sets up a predictable fight: one side will describe the exhibit as a straightforward “system output,” while the other side will describe the same exhibit as a specialized conclusion produced by a process the jury cannot evaluate without a reliability foundation.

The “simple scientific instruments” carveout will sharpen that threshold fight. The Standing Committee agenda book reflects committee discussion about avoiding litigation over the output of instruments that can be presumed reliable while leaving room for judges to decide whether a particular instrument fits within the exception. The exception also signals what the rule is trying to catch: software-driven inference that looks less like measurement and more like conclusion.

Briefing should expect a framing war over purpose. Proponents often prefer a narrow framing that treats output as a record of what a system “did.” Opponents often prefer a purpose framing that treats output as proof of a disputed proposition, which is where reliability arguments become harder to dodge. Committee materials anticipate that dynamic by describing Rule 707 as applying when output is offered directly or accompanied by lay testimony that does not establish reliability.

Real Doesn’t Mean Right

Authentication arguments will still matter, but Rule 707 is designed to prevent authentication from doing reliability’s job. Committee materials emphasize that authenticity and reliability are distinct burdens, and that satisfying a process-based authenticity foundation does not establish admissibility when reliability standards apply. The Standing Committee agenda book describes a key asymmetry: reliability under Rule 702 or proposed Rule 707 is a higher threshold than authenticity under Rule 901, so authentication cannot substitute for reliability.

That distinction becomes practical when a machine output looks polished and authoritative. Authenticity asks whether the proponent has shown the exhibit is what the proponent claims. Reliability asks whether the methods and application justify using the output as proof. Committee materials tie Rule 707 to the second question, warning that a low-threshold authenticity showing does not suffice for admissibility where Rule 702-type reliability concerns exist.

The Evidence Rules Committee has also been working in parallel on deepfakes and authenticity issues, which helps explain why the committee separated the “machine output reliability” problem from the “audiovisual authenticity” problem. The January 2026 Standing Committee agenda book describes two distinct evidentiary challenges: machine-generated evidence that would be subject to Rule 702 if propounded by a human expert, and audiovisual evidence that is not authentic because it is a difficult-to-detect deepfake. Proposed Rule 707 addresses the former concern.

Gatekeeping Gets Granular

Rule 707 would not eliminate fights over machine output. Rule 707 would relocate those fights into predictable pretrial motion practice. The draft tells parties to litigate reliability using the Rule 702 framework even when the proponent does not call a testifying expert. Committee materials describe the rule as regulating machine-generated output pretrial in essentially the same way as expert testimony, while also recognizing that trial presentation may still require careful framing and jury instructions.

Rule 707 also carries a drafting signal that matters to briefing. Committee materials discuss why the rule references Rule 702’s requirements while omitting qualification language that fits humans rather than machines. That choice suggests courts will be asked to translate Rule 702’s familiar concepts into machine-output analysis, including how “helpfulness” works when the “specialized knowledge” lives in software and statistical modeling rather than in a witness.

January 2026 hearing submissions preview the fault lines already forming. One submission argues that the proposal risks functioning as a permissive “pathway” rather than a true gate because the draft uses “may admit,” and because the structure may invite checklist compliance. That critique will likely become a common opposition theme for litigators who want a stricter reliability barrier, and a common defense theme for litigators who want flexibility.

Discovery Targets Reliability Documentation

Rule 707 will not rewrite discovery rules, but reliability burdens tend to create discovery gravity. A party that wants to offer machine output as proof will need a record that supports reliability. A party that wants to challenge machine output will need access to enough information to test reliability. Committee materials acknowledge the practical pressure point: meeting Rule 707 reliability standards may be difficult, and sometimes impossible, without expert testimony and supporting information about how output was generated.

Trade-secret claims and protective-order negotiations will likely intensify around that pressure point. Reliability showings tend to implicate documentation that vendors and deploying parties treat as sensitive: version history, configuration settings, evaluation materials, validation testing results, update cadence, and operational controls. Committee materials keep discovery mechanics outside the Evidence Rules lane, which effectively pushes these disputes into the usual arenas: proportionality, staged production, restricted access, and expert-only review under protective order.

A practical planning takeaway follows from the committee’s own framing. When a proponent plans to rely on machine output, early case assessment should include a reliability-file plan that identifies what records exist, who controls them, and whether vendors can supply supporting materials on a litigation timeline. When an opponent anticipates machine output, early discovery planning should treat reliability as a live issue rather than a late-stage surprise.

Expert Decisions Carry Strategic Weight

Proposed Rule 707 does not mandate experts, but the committee’s own record points to an unavoidable reality: meeting Rule 702-style reliability requirements without a testifying expert can be hard. The Standing Committee agenda book reflects the committee’s expectation that reliability standards will often be difficult to meet without expert testimony, particularly where output depends on specialized methods and a jury cannot evaluate reliability unaided.

That reality turns expert decisions into early strategic choices. A proponent can attempt to admit machine output without an expert, but Rule 707 would still require satisfying the Rule 702 requirements. An opponent can force the question: whether the proponent has enough evidence to establish reliability, or whether the absence of an expert leaves the court with too little foundation to admit the output as proof. Committee materials describe this issue directly by warning against using machine output to evade Rule 702 requirements that would apply if the conclusion came from a human expert.

Jury handling also shows up in committee discussion. The Standing Committee materials describe a tentatively approved committee note concept endorsing a limiting instruction that machine-generated evidence is subject to error and should not be assumed reliable or unreliable simply because a machine produced it. That concept matters because it frames how courts may try to manage the “machine authority” effect even after a reliability ruling.

Governance Effects Extend Beyond Federal Courts

Federal Rules of Evidence amendments apply in federal court, but the governance effects extend beyond federal dockets. Committee materials frame Rule 707 as a response to reliability concerns created by AI and software output offered as evidence, which is a reality for multinational organizations that generate machine output across global systems. When litigation teams need validation records, change logs, and operational documentation to support admissibility, vendors and deploying organizations face pressure to maintain audit-ready reliability artifacts as a routine governance practice, not a special litigation project.

Cross-border disputes add a second layer. Arbitration and investigations often import U.S.-style reliability arguments even when formal evidence rules do not apply, especially when a party wants to challenge the weight of machine output. A U.S. federal rule that treats machine output like expert opinion will also influence how lawyers negotiate vendor contracts and internal policies, as “admissibility readiness” becomes a concrete risk-control reason to demand documentation, testing records, and change-control discipline.

Practice Takeaways for Case Teams

Start with the trigger question. Ask whether the output would be expert testimony if a witness testified to the same conclusion, then frame briefing around that characterization.

Separate authenticity from reliability. Authentication under Rule 901 can establish what a system produced, while proposed Rule 707 targets whether the output is reliable enough to be admitted as proof.

Plan for reliability discovery. A reliability showing usually requires records beyond a printout, including how the system was configured and whether the proponent can support claims of reliable operation.

Decide early whether expert testimony is unavoidable. Committee materials anticipate that meeting reliability standards may be difficult without expert testimony, which makes the expert decision a timing issue, not a trial issue.

Use the rulemaking record as briefing intelligence. Hearing submissions and committee discussion preview arguments that will likely reappear in motions and evidentiary hearings.

Public Comment Window Closes Feb. 16

Rule 707 remains a proposal. The public comment period closes Feb. 16, 2026. Written comments may be submitted electronically through the Regulations.gov docket.

The January 2026 Standing Committee agenda book provides the committee’s most detailed explanation of Rule 707’s purpose, anticipated concerns, and supporting evidence. That record is valuable for practitioners because the committee’s reasoning and framing will likely reappear in judicial opinions and motion practice. The agenda book and other materials are available on the Judiciary’s proposed amendments page.

Public hearings on Rule 707 are scheduled for Jan. 15 and Jan. 29, 2026. Requests to testify must be submitted at least 30 days before the hearing date to the Secretary of the Committee on Rules of Practice and Procedure.

For lawyers who want to follow the proposal as it develops, the complete rulemaking record is maintained on Regulations.gov, while official committee materials and agenda books are available through the U.S. Courts proposed amendments page.

Sources

• Federal Register: Advisory Committees on Appellate, Bankruptcy, Civil, Criminal, and Evidence Rules; Notice of Proposed Amendments and Open Hearings (July 14, 2025)
• Legal Information Institute, Cornell Law School: Federal Rule of Evidence 702, Testimony by Expert Witnesses
• Legal Information Institute, Cornell Law School: Federal Rule of Evidence 901, Authenticating or Identifying Evidence
• Regulations.gov: Proposed Amendments to the Federal Rules of Evidence (docket USC-RULES-EV-2025-0034) (posted Aug. 15, 2025; comment period ends Feb. 16, 2026)
• U.S. Courts: Advisory Committee on Evidence Rules, Agenda Book (Nov. 5, 2025)
• U.S. Courts: Advisory Committee on Evidence Rules, Jan. 15, 2026 Public Hearing Schedule and Testimony Packet (Jan. 2026)
• U.S. Courts: Committee on Rules of Practice and Procedure, Standing Committee Agenda Book (Jan. 6, 2026)
• U.S. Courts: Preliminary Draft of Proposed Amendments to the Federal Rules (Aug. 2025)
• U.S. Courts: Proposed Amendments Published for Public Comment (comment period Aug. 15, 2025 to Feb. 16, 2026)

This article is provided for informational purposes only and does not constitute legal advice. Readers should consult qualified counsel for guidance on specific legal or compliance matters.

See also: Should AI Receive Attorney-Client Privilege or Is Every Prompt a Potential Waiver?