Vermont: AI Ethics Guidance for Law Firms

Bottom line for a 5-50 attorney Vermont firm: No Vermont state court or District of Vermont rule or standing order requires AI disclosure as of 2026-09-30. Accountability runs through the existing Vermont Rules of Professional Conduct, V.R.C.P. 11, and Vermont Supreme Court Administrative Directive No. 38, under which all court users are “expected to review and confirm the accuracy of AI output.” In LaRoche v. Sterett the Vermont Supreme Court found a Rule 11(b)(2) violation where an attorney filed a brief with a fabricated quotation that came from his client’s AI-assisted draft, and it required him to file the order in every pending Superior Court case in which he had appeared. Vermont trial courts have also sanctioned self-represented litigants who cited fabricated authority.

Firms documenting compliance may wish to start with: a written AI use policy, a citation and quotation verification step before any court filing, and a vendor review record covering each AI tool’s data retention and training terms.

What governs by default

The Vermont Bar has issued no formal ethics opinion on AI. Vermont Bar Counsel Michael Kennedy has stated in his unofficial Ethical Grounds blog (2024-05-16) that amending the Vermont Rules of Professional Conduct is unnecessary because existing rules already cover AI use. The Vermont Rules of Professional Conduct named in the page-level applicable-rules block apply directly. Rule 1.1 Comment [8], effective December 10, 2018, states that a lawyer should keep abreast of changes in the law and its practice, “including the benefits and risks associated with relevant technology” (promulgation order). No amendment adding AI-specific language appears in the Judiciary’s Memos to the Bar or promulgated rules as of 2026-09-30.

The First Annual Report (2025-03-01) of the Vermont Judiciary Committee on Artificial Intelligence and the Courts, a committee the Vermont Supreme Court established in 2024, is the principal interpretive document. Bar Counsel Kennedy chaired the Subcommittee on Disciplinary Rules. The subcommittee found that “the rules establish the contours of conduct that is either authorized or prohibited, and that a lawyer’s or judge’s use or misuse of GAI falls within the confines of those contours.” It concluded that “amendments to the rules are not warranted as a result, but that guidance for the bar and for judges is appropriate.”

The report’s guidance addresses specific rules. Rule 1.1: the duty of competence requires lawyers “to be knowledgeable about [GAI’s] risks and benefits.” Rule 1.4: a lawyer “might consider disclosing to a client the lawyer’s intent to use GAI, along with the nature and extent of the use.” Rule 1.6: “a lawyer should understand that there is a risk that the use of GAI can result in an unauthorized disclosure of confidential information,” with guidance on “careful review of a GAI tool’s Terms of Service.” Rule 1.5: “the use or non-use of GAI can have a direct impact on the amount of time spent on a matter,” with different billing considerations described. The report’s “Miscellaneous” section also covers “Meritorious Claims and Contentions,” “Complying with Court Orders,” “Supervision of Others,” “Using GAI at the Direction of a Supervisor,” and “Pro Bono.” The report is non-binding guidance, not a rule amendment or a formal ethics opinion.

The Vermont Supreme Court has since adopted Administrative Directive No. 38, Policy on Artificial Intelligence (amended and effective October 6, 2025). It states that “All court users, including self-represented litigants, are expected to review and confirm the accuracy of AI output and related information provided to the Judiciary or used in connection with a court proceeding,” and that the Vermont Rules of Professional Conduct and the Code of Judicial Conduct “apply fully to the use of AI systems and output.”

No Vermont adoption of ABA Formal Opinion 512 (July 2024) has been located. Bar Counsel has continued publishing on AI competence and confidentiality through Ethical Grounds, including Competence, AI, and the Attorney-Client Privilege (2026-02-17).

What federal courts in Vermont require for AI use in filings

As of 2026-09-30, no District of Vermont general order, local rule, or judge’s page requires AI certification or disclosure. The court’s General Orders (through No. 130) and its Local Rules effective April 20, 2026 contain no AI provision, and the U.S. Bankruptcy Court for the District of Vermont has no AI standing order. Firms documenting compliance may wish to check the assigned judge’s orders at the opening of each new matter. The operative federal default is Fed. R. Civ. P. 11: factual contentions must have evidentiary support, and legal contentions must be warranted.

What Vermont state courts require for AI use in filings

No Vermont state court rule or standing order requires disclosure of AI use in filings. Administrative Directive No. 38 sets an expectation, not a disclosure duty: court users are “expected to review and confirm the accuracy of AI output.” The Judiciary’s proposed and promulgated rule amendments and its Memos to the Bar through September 23, 2026 include no AI-specific rule. V.R.C.P. 11(b) and V.R.A.P. 25(d) apply to every filing, and Vermont courts have applied them to fabricated citations.

Has anyone been sanctioned for AI use in Vermont?

Yes. In LaRoche v. Sterett (Vt. Nov. 13, 2025), the Vermont Supreme Court concluded that Attorney Donald P. LoCascio “violated Vermont Rule of Civil Procedure 11(b)(2) by failing to make a reasonable inquiry into the accuracy of a legal contention contained in a brief he filed with this Court” (entry order). The brief attributed a quotation to Sprague v. Nally that the case does not contain; at oral argument he explained that his client had used an “AI helper” to prepare an initial draft. The Court imposed no monetary sanction but directed him “to file a copy of this entry order into all cases pending in Vermont Superior Court in which he has entered an appearance.” Vermont trial courts have also sanctioned self-represented litigants for fabricated authority, including a fee award and a pre-filing injunction in Lafayette v. Abrami (Dec. 18, 2025).

Separately, as reported by VTDigger (2026-03-01), Rutland attorney Thomas Lamar Enzor used AI in spring 2025 while representing someone charged with drunken driving in Bennington County. Judge Jennifer Barrett “pointed out five mistakes” in the filing and allowed a supplemental motion to correct them, and Lamar Enzor said he did not face other consequences. The Professional Responsibility Board’s public decisions list (latest entry December 17, 2025) includes no AI-related matter.

No Vermont attorney has been publicly disciplined by the Professional Responsibility Board for AI misuse. Elsewhere, People v. Crabill (Colo. 2023) resulted in a suspension of one year and one day, with ninety days served.

What AI-related rules are pending in Vermont?

No Vermont legislation addressing attorney or law firm AI use has been located. H.341 and H.340, introduced 2025-02-25, were referred to the House Committee on Commerce and Economic Development with no recorded action after referral, and H.710 (2024) did not advance past referral. S.23 (synthetic media in elections) was signed on March 5, 2026 as Act 75 and is limited to election communications. Two 2026 acts address AI in health care: Act 101 (H.814, neurological rights and AI in health and human services) and Act 156 (H.816, AI in mental health services). Act 132 (2022) governs State government AI use only. None imposes obligations on law firm practice.

No Vermont bar or Professional Responsibility Board AI opinion has been announced. The Judiciary Committee’s charge calls for annual reports beginning March 1, 2025; as of 2026-09-30 only the First Annual Report is posted on the committee’s page.

How do malpractice carriers in Vermont treat AI use?

ALPS is the Vermont Bar Association’s endorsed professional liability carrier (VBA member benefits). ALPS’s published guidance (August 2025) states that LPL policies “typically do not exclude coverage for claims alleging negligence arising as a result of the use of generative AI,” while cautioning that coverage may depend on whether the conduct meets the policy’s definition of “professional services.” The ABA Journal’s 2025 coverage piece raised a similar caution. No filed AI exclusion has been located in the admitted lawyers professional liability market. Most LPL applications reviewed for this site do not ask about AI; one exception is AmTrust’s application form LPLPRO-APP-01 (05/23), which includes a checkbox for allowing “the use of Artificial Intelligence software to draft documents.” Broker commentary reports LPL underwriters focusing on firms’ AI controls.

What might a Vermont malpractice carrier ask about AI at renewal?

ALPS has not published AI-specific application questions for Vermont, and this site has not located any on its application. Firms documenting compliance may wish to keep five items ready in case a carrier or broker asks: a written AI use policy, a citation verification protocol, vendor review records covering data retention and training terms, attorney and staff training logs, and written confirmation from ALPS or the broker of how AI-related errors are treated under the current policy. ALPS Vermont or the firm’s broker can confirm current application items.

What documentation should a Vermont firm keep on file?

Month one (foundational)

Three foundational items.

1. (Owner: managing partner + firm administrator) Written AI use policy (Rules 5.1, 5.3; Judiciary Committee Report). Enumerate approved tools. Prohibit submitting unverified AI output to courts or clients. Prohibit inputting client-identifying information into consumer-tier tools without enterprise terms. Assign verification responsibility per matter.
2. (Owner: litigation lead) Citation verification checklist for court filings (Rules 3.1, 3.3; V.R.C.P. 11; Fed. R. Civ. P. 11). Pre-filing step requiring independent verification of every AI-generated citation and quotation against Westlaw, Lexis, or the primary source. LaRoche v. Sterett, where the fabricated quotation came from a client’s AI-assisted draft, shows why verification covers material that originates outside the firm.
3. (Owner: firm administrator) Attorney and staff training log (Rules 1.1, 5.1, 5.3). Date, attendees, and content of AI training: hallucination risks, Rule 3.3 candor obligations, verification procedures, confidentiality under Rule 1.6, and the firm’s AI use policy.

Months two and three (operational documentation)

Three documentation items.

1. (Owner: firm administrator + outside IT) Vendor review record (Rule 1.6). For each AI platform: documented review of terms of service, training-data use, opt-out status, data retention, and whether an enterprise agreement prohibits training on inputs. Update when vendor terms change.
2. (Owner: managing partner) Supervision protocol (Rules 5.1, 5.3). Approved tools, prohibited uses, required attorney review before AI-generated work product reaches a court or client, and training requirements for associates, paralegals, and legal assistants.
3. (Owner: billing partner) Billing policy for AI efficiency (Rule 1.5). Written policy addressing how AI-generated time savings are handled for hourly clients. Communicate to billing attorneys. The Judiciary Committee Report notes that “the use or non-use of GAI can have a direct impact on the amount of time spent on a matter,” and outlines billing considerations under Rule 1.5’s prohibition on unreasonable fees.

Months four to six (per-matter discipline)

These are recurring practices, not one-time projects. Each new matter exercises them.

1. (Owner: matter lead attorney) Per-matter AI use notes (Rules 1.1, 3.1, 3.3). Brief attorney note on any matter where AI generated research, drafted pleadings, or produced filed content. Capture which tool, what tasks, and how output was verified.
2. (Owner: matter lead attorney) Client confidentiality consent and engagement letter language (Rules 1.4, 1.6). Engagement letter provision disclosing categories of AI use and confidentiality safeguards. Document written informed consent before client-identifying information enters a third-party AI platform.
3. (Owner: matter lead attorney) Judge-specific AI order check (Rule 3.3; Fed. R. Civ. P. 11). At each new federal matter in D. Vt., review the assigned judge’s individual standing orders and chambers practices for AI disclosure requirements. Document the check.

We are building practitioner resources for firms working through this list. The newsletter covers new resources as they ship.