Does this order require attorneys to disclose AI use?

The order does not require disclosure or certification, but cautions counsel that Rule 11 (or analogous duties) applies to unverified AI content.

What are the consequences for attorneys who violate this order?

The order specifies sanctions, fees, or duties imposed on counsel personally for noncompliance, which may include disclosure to opposing counsel, mandatory CLE, monetary sanctions, or other measures at the court's discretion.

EOIR (Immigration Courts and Board of Immigration Appeals; nationwide): EOIR Policy Memor…

Sirce E. Owen, Acting Director (signed by digital signature, August 8, 2025) · Executive Office for Immigration Review (DOJ)

active

Verified April 28, 2026

Citation: EOIR Policy Memorandum 25-40 (OOD): Use of Generative Artificial Intelligence in EOIR Proceedings
Order date: August 8, 2025

Summary

EOIR has neither a blanket prohibition on the use of generative AI in its proceedings nor a mandatory disclosure requirement regarding its use.

What does the order require?

EOIR has neither a blanket prohibition on the use of generative AI in its proceedings nor a mandatory disclosure requirement regarding its use.
Individual adjudicators or courts may adopt standing orders (per PM 20-09) or local operating procedures (per 8 CFR § 1003.40) regulating the use and disclosure of generative AI in pleadings; the BIA may prescribe its own rules with Director approval under 8 CFR § 1003.1(d)(4).
Parties and attorneys 'should take care to confirm the accuracy of any citations or other research or drafting conducted using generative AI tools.'
Submission of hallucinated or erroneous AI-generated content 'likely violate[s] professional conduct obligations' and may subject attorneys to discipline under 8 CFR § 1003.102(c) (knowingly or with reckless disregard offering false evidence) and §§ 1003.102(k)-(o), (s), (u) (ineffective assistance, prejudice to administration of justice, failure to provide competent representation, failure to disclose adverse legal authority, repeated boilerplate filings reflecting substantial failure to competently and diligently represent client).
Adjudicators who suspect or discover inappropriate AI use should report those instances to EOIR's Attorney Discipline Program and, as appropriate, its Anti-Fraud Program.
The PM does not create any right or benefit enforceable at law or in equity by any party against the United States; it is general guidance for adjudicators.

Practice areas: immigration

Verify this order against the court's official website before relying on it. Standing orders are amended without notice. Requirements vary by judge and case type.

What the memorandum says

EOIR Policy Memorandum 25-40, signed by Acting Director Sirce E. Owen on August 8, 2025 (effective same date), is the first DOJ-component AI policy binding immigration practitioners. It applies to all EOIR adjudicatory components: Immigration Courts and the BIA, nationwide.

The PM takes a deliberate non-prescriptive posture. It does not require disclosure, does not impose a separate certification, and does not categorically ban AI use. Its operative requirements run through three mechanisms:

Verification duty. Parties and attorneys “should take care to confirm the accuracy of any citations or other research or drafting conducted using generative AI tools.”
Professional-conduct discipline. Hallucinated AI content “likely violate[s] professional conduct obligations” and exposes counsel to discipline under 8 CFR §§ 1003.102(c), (k)-(o), (s), (u). The list is broad: false evidence, ineffective assistance, prejudice to the administration of justice, failure to provide competent representation, failure to disclose adverse authority, and repeated boilerplate filings.
Delegated standing-order authority. Individual immigration judges, BIA panels, and individual courts may adopt their own AI standing orders or local operating procedures. Practitioners must therefore check the assigned IJ’s standing orders in addition to the PM.

Why this matters for immigration practice

EOIR is the largest administrative court system by case volume in the federal government. The PM acknowledges EOIR “has largely lagged behind” other courts on AI guidance and signals that future rulemaking and a possible BIA precedential decision are anticipated. Until then, the PM is the operative DOJ-component framework for immigration filings.

The PM is also “largely consistent with OCAHO’s established guidance” (the Office of the Chief Administrative Hearing Officer issued United States v. Wallcon, LLC, 21 OCAHO no. 1630, 9-14 (2025), addressing AI use through adjudication). Where the PM and OCAHO conflict, OCAHO controls in OCAHO matters.

Practitioner workflow

Immigration matters before any IJ or the BIA: no separate disclosure or certification required by the PM itself, but check chambers-level standing orders (the PM expressly authorizes them) and verify any AI-generated citation before filing. The discipline exposure is broad (eight separate subsections of 8 CFR § 1003.102 are flagged), so a hallucinated cite is a discipline question, not just a Rule 11-equivalent question.

The PM also warns about future DOJ-wide guidance: OMB Memorandum M-25-21 required each federal agency to develop generative-AI policy by approximately December 29, 2025. DOJ-wide guidance was expected by year-end 2025; immigration practitioners should track for an update that may supersede or extend PM 25-40.

Primary source

PM 25-40 PDF: https://www.justice.gov/eoir/media/1410621/dl?inline