By Vinitha Prasannan, Senior Manager – Delivery
We live in an age of rapid technological change. Society has moved beyond the digital era into a world shaped by automation, artificial intelligence, generative AI, agentic AI, and related technologies. Courtrooms now routinely encounter digital evidence including emails, text messages, social media posts, metadata, surveillance footage, and AI-generated content as a part of litigation.
Traditionally, courts have relied on evidence such as oral testimony, documentary evidence, and physical evidence. Digital evidence, however, raises new and complex questions concerning authenticity, reliability, accuracy, preservation, and admissibility. The National Institute of Justice (NIJ) defines digital evidence as “information stored or transmitted in binary form that may be relied on in court.”[1]
Reports indicate that more than 90% of criminal cases now involve some form of digital evidence.[2] According to the NIJ, law enforcement agencies are integrating digital-forensic expertise into routine criminal investigations because such evidence is relevant to nearly every type of crime.[3] This development requires agencies to train personnel in the proper preservation and collection of digital evidence while simultaneously adapting evolving operating systems and other technologies.
As digital evidence becomes increasingly central to modern litigation, corresponding changes in the legal system are not merely desirable — they are inevitable. The Federal Rules of Evidence (FRE) are not static; they are regularly revised to address emerging legal and technological challenges.
The Foundational Rules Governing Digital Evidence
Electronically stored information may help prove or disprove facts in a legal proceeding, but such evidence is fragile and may be altered if handled improperly. Investigators must therefore follow strict procedures when identifying, collecting, acquiring, transferring, and preserving digital evidence. Its admissibility may depend on lawful acquisition — through a warrant, consent, or exigent circumstances — and on establishing its authenticity and reliability through a properly documented chain of custody. In federal courts, the Federal Rule of Evidence primarily govern admissibility, while state courts generally apply comparable state evidentiary rules.
Several foundational rules work together to govern the admission of digital evidence. Evidence must be relevant under Rules 401 and 402, but a court may still exclude it under Rule 403 if its probative value is substantially outweighed by the risk of unfair prejudice, confusion, delay or other specified concerns. Under Rule 602, witness testimony must ordinarily be based on personal knowledge. Out-of-court statements may implicate the hearsay rules and the applicable exclusions or exceptions under Rules 801 through 804.
A party must properly authenticate documentary, electronic, and recorded evidence under Rule 901. Rule 1002 generally requires an original when a party seeks to prove the contents of a writing, recording, or photograph, subject to the exceptions and alternatives permitted by the other rules in Article X. Additional provisions, including Rules 702 and 404, govern expert testimony and character evidence, respectively.
Authentication Under Rule 901
Federal Rule of Evidence 901 is the primary rule governing the authentication or identification of evidence. It requires the proponent to produce sufficient evidence to support a finding that the item is what the proponent claims it to be. Authentication may be established through witness testimony, handwriting or voice identification, distinctive characteristics, comparison with an authenticated specimen, evidence concerning public records, telephone conversations, or proof that a process or system produces an accurate result. The methods listed in Rule 901 are illustrative rather than exhaustive, allowing courts to consider other reliable methods of authentication permitted by law.
Rule 902(13) permits records generated by an electronic process or system to be authenticated through a certification by a qualified person showing that the process or system produces an accurate result. The provision allows digital evidence — such as computer logs, automated records, and data generated by electronic systems — to be admitted without requiring a technical witness to testify solely for authentication purposes. The opposing party must, however, receive reasonable written notice and an opportunity to inspect the records and certification.
Even with these streamlined authentication procedures, some commentators argue that traditional evidentiary doctrines, particularly the hearsay rules and their numerous exceptions, remain unnecessarily complex.
FRE 702 and the Rise of Machine-Generated Evidence
The amendment to Federal Rule of Evidence 702, effective December 2023, clarified and reinforced the court’s gatekeeping obligation. The amended rule requires the proponent of expert testimony to establish, by a preponderance of the evidence, that each admissibility requirement is satisfied. It also requires the court to determine that the expert’s opinion reflects a reliable application of reliable principles and methods to the facts of the case.
At the same time, the increasing use of deepfakes, artificial intelligence, and machine-generated data has prompted consideration of specialized evidentiary provisions, including proposed Rule 707. The proposed rule would extend Rule 702’s reliability requirements to certain machine-generated evidence offered without an expert witness. Under the proposal, when machine-generated evidence would be subject to 702 if presented as a human expert testimony, a court could admit the evidence only if it satisfies the requirements of Rule 702(a) through (d). As of August 2026, however, Rule 707 remains under consideration and has not been adopted as part of the Federal Rules of Evidence.
The proposed rule could apply to outputs involving matters such as algorithmic analyses of stock-trading patterns, comparisons of computer software, or assessments of possible source-code misappropriation. Even when a party introduces such evidence through a lay witness or authenticates it through a certification under Rule 902(13), the party would still need to establish the reliability of the machine-generated output when proposed Rule 707 applies. Authentication establishes that the evidence is what its proponent claims and it does not necessarily establish that the system’s analysis or conclusions are reliable.
Until a specialized rule is adopted, federal courts must evaluate algorithmic and machine-generated evidence under the existing rules governing relevance, authentication, expert testimony, and reliability. Under Rule 901 Under Rule 901(b)(9), a proponent may authenticate evidence by describing the process or system that produced it and showing that the process or system produces an accurate result. Rules 902(13) and 902(14)
permit certain electronically generated records and electronically copied data to be authenticated through certifications from qualified persons, subject to applicable notice requirements. Rule 902(14), however, concerns the authenticity of data copied from an electronic device, storage medium, or file and it does not independently establish the reliability of an algorithmic conclusion.
When evidence depends on complex scientific or technical methods, courts may apply Rule 702 and the Daubert framework[4] to examine factors such as whether the methodology can be and has been tested, its known or potential error rate, the existence of standards governing its operation, peer review and publication, and its level of acceptance within the relevant field. These considerations are flexible rather than an inflexible checklist.
Purely machine-generated output ordinarily does not constitute hearsay, when it does not communicate a human assertion because a machine is not a “declarant” within the meaning of Rule 801. Human statements, assumptions, labels, or data incorporated into or surrounding the automated process may nevertheless present separate hearsay concerns. Proprietary “black box” systems, deepfakes, and generative AI have therefore intensified demands for transparency, access to underlying methodologies, adequate disclosure, and stronger procedures for establishing authenticity and reliability.
The Advisory Committee’s 2025 Proposals
The Advisory Committee on Evidence Rules’ May 2, 2025 proposals acknowledge that existing evidentiary rules may not adequately address evidence generated by artificial intelligence.[5] In its proposal, the Committee identified two principal concerns: first, the reliability of machine-learning outputs, and second, the authenticity of audiovisual material, particularly deepfakes. Because it is impossible to test machine-generated evidence through cross-examination in the same manner as human testimony, the Committee concluded that the reliability of such evidence raises concerns similar to those already governed by Rule 702. Because Rule 702 was revised in 2023 and applies broadly to expert testimony, the Advisory Committee proposed Rule 707 specifically to address machine-generated evidence.
The proposed amendments to Rule 901 would also address the growing threat of deepfakes and other forms of synthetic media. Under the proposal, when a party makes a credible, evidence-based claim that an image, video, or audio recording has been fabricated or materially altered by artificial intelligence, the court may require stronger proof of authenticity before admitting it. This would give judges a more active gatekeeping role, including the ability to examine the digital chain of custody, metadata, watermarking, or other technical indicators of authenticity.
The Committee further concluded that detailed disclosure requirements for machine-generated evidence would be better addressed through the Civil and Criminal Rules rather than the Evidence Rules, although it emphasized the importance of advance notice. The proposal was published for public comment, and the comment period ended on February 16, 2026. The comments were expected to help determine whether proposed Rule 707 was necessary, what types of machine-generated evidence it should cover, and which technologies should be exempt. Given the rapid pace of AI development and the Committee’s limited technical expertise, input from legal and technological experts was expected to play a central role in shaping any final rule. In May 2026, however, the Committee declined to recommend further action on proposed Rule 707 at that time.[6]
Looking Ahead
As digital evidence continues to grow more sophisticated — from routine metadata to fully AI-generated content — the rules governing its admissibility will need to keep pace. Practitioners should continue to monitor developments concerning proposed Rule 707 and possible amendments addressing authentication because they may eventually reshape how courts evaluate the reliability and authenticity of machine-generated evidence in both civil and criminal proceedings
[1] https://nij.ojp.gov/topics/forensics/digital-multimedia-evidence
[2] https://pmc.ncbi.nlm.nih.gov/articles/PMC10311201/#bib1
[3] https://nij.ojp.gov/topics/forensics/digital-multimedia-evidence
[4] a systematic framework articulated in Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), used by federal court judges to assess the reliability and relevance of expert witness testimony before presenting it to a jury.
[5] https://www.uscourts.gov/sites/default/files/document/preliminary_draft_of_proposed_amendments_final.pdf
[6] https://www.uscourts.gov/sites/default/files/document/advisory_committee_on_evidence_rules_may_2026.pdf
For more practical litigation drafting tips, pleading and legal research insights, follow LawCompany.