Federal Court AI Standing Orders: Disclosure Requirements Explained

Federal courts began issuing AI-specific standing orders in 2023, and by the end of 2024 the practice had spread to dozens of district courts and several circuit courts. The orders are not uniform. Some require only disclosure that AI was used. Others require a certification that an attorney reviewed every AI-generated passage for accuracy. A handful prohibit AI-generated content in filings entirely, at least without prior leave. A few courts have issued no order at all.

For litigators, the practical problem is that the landscape is fragmented by judge and by district. A standing order from one judge in a district does not bind other judges in the same courthouse. That means checking the assigned judge's individual standing orders — not just the local rules — before every filing.

Why Courts Started Issuing These Orders

The proximate cause was a documented pattern of AI-generated citation hallucinations appearing in federal court filings. The most widely cited early incident involved fabricated case citations submitted in the Southern District of New York in 2023, resulting in sanctions against the attorneys involved. That case drew judicial attention across the federal system to the fact that existing rules — Rule 11 certification, professional responsibility obligations, and general competence requirements — did not explicitly address AI use, and that attorneys were not uniformly treating AI output with the same verification scrutiny they applied to other research tools.

Standing orders were the fastest mechanism available to individual judges who wanted to address the issue without waiting for formal local rule amendments, which require public comment periods and approval processes that can take years.

The Four Requirement Types You Will Encounter

Across the federal courts that have issued AI-related orders, requirements cluster into four categories. Understanding which type applies to your case determines what you need to include in your filing.

How Disclosure Orders Are Structured

Disclosure-only orders are the most common type. They typically require a short statement — often a single sentence or a checkbox-style certification — appended to or included within the filing. The exact form varies by order. Some judges specify the precise language; others describe the substance and leave the form to the attorney.

A disclosure statement under these orders generally needs to cover:

• Whether any generative AI tool was used in drafting or researching the document
• Which tool or tools were used (some orders require naming the tool; others require only a yes/no)
• Whether the attorney reviewed the AI-generated content for accuracy
• A statement that the attorney takes responsibility for the content regardless of AI involvement

Some orders draw a distinction between generative AI tools (which require disclosure) and AI-assisted legal research platforms like Westlaw or Lexis (which may not, depending on the order's language). That distinction matters in practice: a litigator using Westlaw's AI-assisted research features may not need to disclose that use under an order that is scoped to "generative AI" or "large language models," but should read the specific order carefully before assuming.

Certification Orders: What They Add

Certification orders go further than disclosure. They require the attorney to affirmatively certify — under their professional responsibility obligations — that AI-generated content was reviewed and verified. The practical effect is to make explicit what Rule 11 already implies: that the filing attorney personally stands behind the accuracy of every citation and legal assertion in the document.

Several certification orders specifically require that all cited cases actually exist, that the quoted language appears in the cited source, and that the cited authority supports the proposition for which it is cited. Those requirements are not novel — they reflect existing professional responsibility obligations — but making them explicit in a standing order creates a clearer basis for sanctions when they are violated.

Which Courts Have Issued Orders: A Partial Reference

The following table records a selection of federal courts and judges with documented AI-related standing orders or local rules as of mid-2026. This is not a complete registry. Orders are issued at the judge level as well as the court level, and the population changes as new orders are issued and existing ones are amended.

How to Check Before You File

There is no single centralized registry of federal AI standing orders. The practical workflow for compliance involves three steps:

1. Check the court's local rules page. Court-wide AI policies, if any, are typically posted there. Some district courts have added AI-specific provisions to their general filing requirements.
2. Check the assigned judge's individual standing orders page. Most federal district court websites have a page for each judge listing their standing orders as downloadable PDFs. These pages are not always updated in real time, so also check whether the judge has issued any recent orders in the specific case via PACER.
3. Check the case docket for any case-specific orders. Some judges issue AI-related orders at the start of a specific case rather than as a standing order. A scheduling order or initial case management order may contain AI disclosure requirements.

For circuit court filings, check the court's published rules and any general orders posted on the court's website. Both the Fifth and Seventh Circuits had issued court-wide AI disclosure requirements by 2024, and other circuits have followed or are considering similar measures.

What Happens When You Do Not Comply

Noncompliance with an AI standing order has produced a range of judicial responses, from written warnings and orders to show cause, to monetary sanctions, to striking the offending filing. In cases involving fabricated citations — which is a distinct issue from failing to disclose AI use — courts have imposed sanctions under Rule 11, 28 U.S.C. § 1927, and the court's inherent authority.

The more common scenario is not fabrication but failure to disclose. Where an attorney used AI and simply omitted the required disclosure statement, judicial responses have generally been less severe — an order requiring a corrected filing, or a warning — but the outcomes depend heavily on the individual judge and whether the underlying content contained any inaccuracies.

The Relationship Between Standing Orders and Professional Responsibility Rules

Standing orders operate alongside — not instead of — professional responsibility obligations. An attorney who complies with a disclosure-only standing order by appending the required statement has not thereby satisfied their Rule 11 obligations or their duty of competence under Model Rule 1.1. The standing order addresses what must be disclosed to the court; the professional responsibility rules address what the attorney must actually do before signing and filing the document.

Several state bars have issued formal ethics opinions addressing AI use in legal practice, covering competence, confidentiality, and supervision obligations that apply regardless of what any particular court's standing order requires. Those opinions are tracked separately in the bar ethics guidance section of this site.

Supervision of AI Output

Model Rule 5.3 requires attorneys to supervise nonlawyer assistance in a way that ensures compliance with professional conduct rules. Several bar opinions have concluded that AI tools used in legal work qualify as nonlawyer assistance for this purpose, meaning the supervising attorney is responsible for the output as if it were produced by a supervised paralegal or contract attorney.

That framing has practical implications for how AI-generated drafts are reviewed before filing. It is not sufficient to read the output for general plausibility. The attorney needs to verify the specific claims — citations, quoted language, characterizations of holdings — against primary sources before signing the filing.

Confidentiality Considerations

Uploading client documents or case-specific facts to an AI tool raises confidentiality issues under Model Rule 1.6 that are separate from the court's standing order requirements. Before using any AI tool for work on a pending matter, attorneys should verify the tool's data retention and training policies. Tools that retain user inputs for model training create confidentiality risks that disclosure to the court does not resolve.

Drafting a Compliant Disclosure Statement

Where a standing order requires disclosure but does not prescribe exact language, the following elements are generally sufficient to satisfy the disclosure requirement. Always read the specific order first — if it prescribes language, use that language exactly.

• Identify whether AI was used (yes/no)
• If yes, identify the tool or tools used by name
• State that the filing attorney reviewed the AI-generated content
• State that all citations were independently verified against primary sources
• State that the attorney accepts responsibility for the accuracy of the filing's content

Some orders require the disclosure to appear in a specific location — at the end of the brief, in the certificate of service, or as a standalone certification. Placement matters. A disclosure buried in a footnote may not satisfy an order that requires a stand-alone certification.

What These Orders Do Not Cover

AI standing orders generally address filings submitted to the court. They do not, on their face, govern:

• Internal work product that is not filed (drafts, research memos, internal analyses)
• Client communications, even if AI-assisted
• Deposition preparation materials
• Settlement communications
• AI use during oral argument (though some judges have addressed this separately)

That scope limitation does not mean AI use in those contexts is unregulated — professional responsibility rules apply — but it does mean the standing order's specific disclosure and certification requirements are not triggered.

Tracking Changes Over Time

The court standing orders landscape is not static. Orders are issued, amended, and occasionally rescinded. Judges who initially issued prohibition orders have in some cases revised them to disclosure-only requirements as familiarity with AI tools increased. Courts that had no order have added one. The practical implication is that a compliance check done at case inception may be stale by the time a major brief is due months later.

Building a standing order check into the pre-filing workflow — rather than treating it as a one-time onboarding task — is the more reliable practice. For cases with multiple briefing deadlines, check the judge's standing orders page before each major filing.