D. Conn.: Chambers Preferences, Policy on AI Research (Judge Vernon D. Oliver, D. Conn.)
Hon. Vernon D. Oliver, U.S. District Judge · U.S. District Court for the District of Connecticut
Verified September 29, 2026
- Citation
- Chambers Preferences, Policy on AI Research (Judge Vernon D. Oliver, D. Conn.)
- Order date
- 2025
Status and scope
No AI-specific filing obligation: in force, but it imposes no disclosure, verification, or use restriction on attorney filings beyond existing rules. It provides: '[A]ttorneys and pro se litigants alike should exercise great caution in submitting any AI-generated language in filings before the Court.'
What the instrument provides
- '[A]ttorneys and pro se litigants alike should exercise great caution in submitting any AI-generated language in filings before the Court.'
- 'Use of AI without verification of the accuracy of the information it generates', like any other shoddy research method, 'implicates Federal Rule of Civil Procedure 11', which 'applies fully to actions filed by pro se litigants.'
- '[A]ll parties are on notice that the Court has a no-tolerance policy for any briefing (AI-assisted or not) that hallucinates legal propositions or otherwise severely misstates the law. Such filings will often result in sanctions absent reasonable excuse.'
- The policy imposes no disclosure or certification requirement.
Practice areas: federal civil
What the policy says
The “Policy on AI Research” on Judge Vernon D. Oliver’s chambers page acknowledges that attorneys and pro se litigants “have begun to use artificial intelligence (‘AI’) to assist with legal research” and that “AI as a legal research tool is here to stay,” but warns that AI “also presents risks: Namely, ‘hallucinated’ cases or otherwise patently incorrect statements of law.”
It then states that “attorneys and pro se litigants alike should exercise great caution in submitting any AI-generated language in filings before the Court,” and that “Use of AI without verification of the accuracy of the information it generates”, like “any other shoddy research method from other sources or tools”, “implicates Federal Rule of Civil Procedure 11.” Rule 11, it adds, “applies fully to actions filed by pro se litigants.”
The policy closes: “Therefore, all parties are on notice that the Court has a no-tolerance policy for any briefing (AI-assisted or not) that hallucinates legal propositions or otherwise severely misstates the law. Such filings will often result in sanctions absent reasonable excuse.”
Relationship to the district notice
The same two closing sentences appear in the District of Connecticut’s Notice to Counsel and Litigants Regarding AI, issued by the Clerk on September 12, 2025. Judge Oliver’s page carried them by July 2025.
Revisions
The policy’s text has not changed since it first appeared in 2025, but its supporting citations have. The 2025 version cited Willis v. U.S. Bank Trust National Association, No. 3:25-cv-516-BN (N.D. Tex. May 15, 2025) (Standing Order Regarding Use of Artificial Intelligence, ECF 48), a magistrate judge’s standing order on AI entered in that case, whose language the policy partly tracks. The current version cites instead:
- Barteca Holdings LLC v. Tacobarn Newtown LLC, No. 26-CV-250 (VDO) (D. Conn. Aug. 4, 2026) (order imposing sanctions), Judge Oliver’s own decision;
- Andre v. Warden, FCI Danbury, No. 3:24-cv-1295 (VDO) (D. Conn. Nov. 25, 2025), also his;
- Cojom v. Roblen, LLC, No. 3:23-cv-1669 (JCH) (D. Conn. Nov. 17, 2025) (sanctions order); and
- Mata v. Avianca, Inc., No. 22-cv-1461 (PKC), 678 F. Supp. 3d 443 (S.D.N.Y. June 22, 2023) (opinion and order on sanctions).
Each citation was checked against the linked opinion: its caption, docket number, and signature date. The links for Barteca, Andre, Cojom, and Mata are the court-hosted copies the policy itself links. The Willis link is to the order in CourtListener’s RECAP archive.
Primary source
Judge Vernon D. Oliver, U.S. District Court for the District of Connecticut (chambers preferences)