By Bill Hartzer · Published · How this site is researched
Evidence rules, court rules and professional guidance
This part of the record tracks the written rules and guidance that govern the use of artificial intelligence in litigation. It covers rules of evidence, court rules and standing orders, ethics opinions, and guidance written for expert witnesses. Each entry states what a rulemaking body or regulator proposed, adopted or issued, and where the matter stands on the date it was checked. It does not state what a rule means beyond its text. Where a rule is still a proposal, the entry says so, and it records the last formal action taken.
The entries are drawn from the rule text or the body's own publication wherever that document could be found and opened. Where an entry relies on a law firm summary or a news report, it says so, and anything that could not be confirmed is listed separately.
Verification, disclosure and confidentiality
Four patterns recur across the bodies covered here. The first is verification. Nearly every instrument, from the Fifth Circuit's June 2024 statement to New York's Part 161 model rule, places the duty to check citations and content on the person who signs or submits the document. The Fifth Circuit put it directly: "'I used AI' will not be an excuse for an otherwise sanctionable offense."
The second is confidentiality. ABA Formal Opinion 512, California rule 10.430, New York's interim policy and the judicial guidance in England and Wales each address what may be entered into a public AI system. The English guidance states that anything entered into a public chatbot "should be seen as being published to all the world."
The third is disclosure, and here the bodies differ. Judge Brantley Starr in the Northern District of Texas required a certificate in every case. The Fifth Circuit considered a similar certificate and did not adopt it. New York's Part 161 expressly does not require disclosure for court papers, and the judicial guidance in England and Wales states that a legal representative using AI responsibly need not refer to its use. The Civil Justice Council has proposed a different answer for expert reports.
The fourth is the reliability of AI output offered as evidence. That question is the subject of proposed Federal Rule of Evidence 707, which remains under study.
Matters in this topic
7 matters, most recent first within each group. 7 were checked against the primary document, and the others rest on reporting and say so.
Rules of evidence
- Proposed Federal Rule of Evidence 707Primary document readThe federal Advisory Committee on Evidence Rules published proposed Rule 707 for comment in August 2025. In May 2026 it declined to recommend adoption, revised the draft, and set it for further study. It also set aside a deepfake amendment to Rule 901.
Court rules and standing orders
- New York courts: interim AI policy and Part 161Primary document readNew York's Unified Court System issued an interim AI policy for judges and staff in October 2025. By an order of March 25, 2026 it adopted Part 161, effective June 1, 2026, which permits AI in preparing court papers without a disclosure requirement.
- AI Judicial Guidance, England and Wales (October 2025)Primary document readOn October 31, 2025 the senior judiciary of England and Wales issued updated guidance on AI for judicial office holders and their staff, replacing the April 2025 version. It covers confidentiality, accuracy, AI use by lawyers and litigants, and deepfakes.
- California Rule of Court 10.430 and Standard 10.80Primary document readOn July 18, 2025 the Judicial Council of California adopted rule 10.430 and Standard 10.80. Courts that do not prohibit generative AI by staff and judicial officers had to adopt a use policy by December 15, 2025.
- Judge Starr's AI certificate and the Fifth Circuit rulePrimary document readIn May 2023 Judge Brantley Starr (N.D. Tex.) required counsel to certify whether generative AI drafted any filing. The Fifth Circuit proposed a similar certificate in November 2023 and decided in June 2024 not to adopt it.
Ethics opinions
- ABA Formal Opinion 512Primary document readOn July 29, 2024 the ABA Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512, its first opinion applying the Model Rules of Professional Conduct to lawyers' use of generative AI tools.
Expert-facing guidance
- Civil Justice Council consultation on AI in court documentsPrimary document readIn February 2026 the Civil Justice Council of England and Wales consulted on rules for AI use in preparing court documents, including a proposal that expert reports identify and explain AI use. Its mid-2026 update pointed toward transparency for expert evidence.
Most instruments govern lawyers and courts, not experts
Most formal action to date governs lawyers and court personnel. California's rule 10.430 and New York's interim policy govern the courts' own staff and judges. ABA Formal Opinion 512 addresses lawyers under the Model Rules of Professional Conduct. Florida's Supreme Court, in an opinion of May 28, 2026 reported by the Florida Bar News, amended its rule on the effect of a signature so that the signer represents that the legal authorities cited exist and are accurately cited.
Rules of evidence have moved more slowly. The federal Advisory Committee on Evidence Rules published Rule 707 for comment in 2025. In May 2026 it declined to recommend adoption, revised the draft and set it for further study by technology experts at its fall 2026 meeting. The same committee concluded that an amendment to Rule 901 to address deepfakes (fabricated audio, video or images) was not warranted for now. The evidence amendments the federal judiciary published for comment in August 2026 do not include either proposal.
Guidance written specifically for experts has so far come mainly from outside the United States. In England and Wales, the Civil Justice Council has consulted on requiring experts to identify and explain AI use in their reports. Its final report is expected later in 2026.
Where the instruments meet the Standard
Read against the Expert Record Standard, the instruments converge on a small set of obligations. The verification requirements correspond to Clause 3, independent authority, and Clause 4, checkability before adoption: nothing enters the work on a system's authority, and output is used only if it has been checked. The confidentiality provisions correspond to Clause 6. An expert who holds material under a protective order is in at least as constrained a position as the court staff these policies address.
The disclosure provisions are where an expert's exposure is least predictable. Requirements are set by individual judges, by courts and by procedural rulemakers, and they change. Some apply only to briefs. New York's Part 161 expressly excludes material offered as evidence from its definition of papers, leaving expert evidence to other requirements. The Civil Justice Council's proposal would reach the expert's report directly. Clause 8 requires disclosure where rules, orders or the engagement require it. Clause 7 makes disclosure possible, because a contemporaneous record of which system was used, on what material and for what purpose is the only reliable basis for an accurate statement later.
Every instrument collected here that addresses experts at all makes the same point: the opinion must remain the expert's own. That is Clause 2, non-delegation.
No settled national rule on machine output as evidence
The record does not show a settled national rule in the United States on AI output offered as evidence. Rule 707 has not been adopted, and the committee's own report describes the public comments as divided. The record does not show a U.S. state rule, located in the sources reviewed for this section, that requires an expert to disclose AI use in an expert report. It does not show how courts will apply the rules and policies collected here in contested cases, and the entries do not predict that.
The record also does not show that guidance from a professional body binds a court. ABA Formal Opinion 512 and the ethics opinions like it are guidance. They may inform what a court or a cross-examiner expects, but they are not rules of court.
Proposal, recommendation, adoption and effective date
When a new rule, order or opinion on AI in litigation is reported, first identify the body and the stage. A proposal published for comment, a committee recommendation, an adopted rule and an effective rule are different things, and news reports often blur them. The Rule 707 proposal was described in some headlines as approved when it had only been approved for publication.
Next, identify who the instrument governs: lawyers, parties, court staff, judges or experts. Many instruments that are reported as AI rules for litigation govern only the court's internal use or only the papers counsel file. Then check whether the instrument covers evidence, since several expressly do not. Finally, locate the text on the body's own website, check the effective date, and check whether a judge-specific order in the case adds to or departs from the general rule. An entry in this record gives the date on which its status was last checked.
Primary documents behind this topic
The orders, opinions and published texts that entries in this topic were checked against. Read them before relying on any summary, including the ones here.
- New York courts: interim AI policy and Part 161
Administrative Order AO/75/2026 adopting Part 161; UCS Interim Policy on the Use of Artificial Intelligence - Proposed Federal Rule of Evidence 707
Report of the Advisory Committee on Evidence Rules (May 2026) - Civil Justice Council consultation on AI in court documents
Civil Justice Council, Use of AI in preparing court documents (working group page and update) - AI Judicial Guidance, England and Wales (October 2025)
Artificial Intelligence (AI) Guidance for Judicial Office Holders (31 October 2025) - California Rule of Court 10.430 and Standard 10.80
California Rules of Court, rule 10.430 - ABA Formal Opinion 512
ABA Formal Opinion 512, Generative Artificial Intelligence Tools - Judge Starr's AI certificate and the Fifth Circuit rule
Fifth Circuit, Court Decision on Proposed Rule (generative AI)
Questions about this topic
Is there a federal rule of evidence on AI-generated evidence?
No, not as of September 11, 2026. Proposed Rule 707 was published for comment in 2025, but in May 2026 the Advisory Committee on Evidence Rules declined to recommend adoption and revised the draft for further study. The August 2026 package of proposed evidence amendments does not include it.
Do courts require lawyers to disclose the use of AI in filings?
Some do and many do not, and the answer depends on the court and the judge. Judge Brantley Starr in the Northern District of Texas required a certificate in 2023, while the Fifth Circuit decided in 2024 not to adopt one. New York's Part 161, effective June 1, 2026, permits AI use in court papers without a disclosure requirement but offers a model certification rule that individual courts may adopt.
Must an expert witness disclose AI use in an expert report?
It depends on the rules, orders and engagement in the particular case. No U.S. rule requiring it was located in the sources reviewed for this section. In England and Wales, the Civil Justice Council has proposed that expert reports identify and explain AI use,.
Does ABA Formal Opinion 512 apply to expert witnesses?
Not directly. The opinion interprets the ABA Model Rules of Professional Conduct, which govern lawyers. Its guidance on verification and confidentiality is relevant to the lawyers who retain and work with experts, and it parallels obligations experts carry under their own professional duties.
What happened to the proposed deepfake amendment to Rule 901?
The Advisory Committee on Evidence Rules concluded in May 2026 that, at least for now, an amendment to Rule 901 to address deepfakes is not warranted. It cited a Federal Judicial Center survey in which 15 judges reported dealing with deepfake issues, and it planned a mini-conference on the subject for its fall 2026 meeting.