Oregon: AI Ethics Guidance for Law Firms

Bottom line for a 5-50 attorney Oregon firm: OSB Formal Opinion 2025-205 (February 2025) is official advisory guidance with directly actionable verification, billing, confidentiality, and supervision duties under Oregon RPCs 1.1, 1.5, 1.6, 3.3, 4.1, 5.1, and 5.3. Oregon enforcement has been aggressive: a roughly $110,000 federal penalty in Couvrette v. Wisnovsky and a record $10,000 Court of Appeals sanction in Doiban v. OLCC make it one of the highest-risk states for AI-related sanctions. Local counsel who sign AI-assisted filings share independent liability for verification failures.

Start with: a written AI use policy (closed-system default), a citation verification log routed through Westlaw or Lexis, and a vendor due diligence record for each AI tool that touches client information.

OSB Formal Opinion No. 2025-205

Citation: OSB Formal Opinion No. 2025-205, “Artificial Intelligence Tools” (approved February 2025 by the Oregon State Bar Legal Ethics Committee and Board of Governors). Primary source PDF. Incorporates the confidentiality framework of OSB Formal Opinion 2011-188 (cloud computing, revised 2015).

Status: Official formal advisory opinion. Persuasive authority in Oregon disciplinary proceedings; not binding on courts.

What it requires

• Under Oregon RPC 1.1, lawyers must understand the capabilities and limitations of any AI tool before relying on its output, with due diligence on the tool before use. The opinion cites Mata v. Avianca, 678 F. Supp. 3d 443 (S.D.N.Y. 2023) as the cautionary case.
• Under Oregon RPC 1.6, before using any AI tool with client information the lawyer must determine whether client consent is required (presumptively yes for open or consumer models). Vendor contracts must also be reviewed for confidentiality parameters, and the lawyer must confirm whether the model retains prompts for training. Risks, benefits, and alternatives must be clearly described to the client.
• Under Oregon RPCs 3.3 and 4.1, “lawyers must review for accuracy any [generative AI] output discussing case-specific facts or providing a case citation, quotation, or conclusion” before submission to a tribunal or third party.
• Under Oregon RPC 1.5, lawyers must inform clients in writing before charging for actual AI tool costs. Hours must not be inflated to recapture time saved by AI.
• Under Oregon RPCs 5.1 and 5.3, supervising and managing lawyers are responsible for ensuring subordinates and non-lawyer staff use AI in compliance with the RPCs. The responsibility is non-delegable.

What it recommends

• Closed systems using only internal data are the safer default. The PLF separately advises that lawyers should not use any AI program unless it is a closed system.
• Clients who object to AI use should be offered alternatives.

Notable gaps

The opinion does not impose a blanket AI-use disclosure requirement to courts (contrast Florida and Texas). It does not address client-facing AI tools, provides no model consent form or policy template, and sets no minimum competency benchmarks for specific AI tools.

What federal courts in Oregon require for AI use in filings

As of April 2026, the U.S. District Court for the District of Oregon has no district-wide standing order or local rule specifically governing AI use in filings. The operative federal default is Fed. R. Civ. P. 11, paired with Oregon RPC 3.3. A court conference panel on AI use was scheduled for May 7, 2026, and may precede a standing order.

What Oregon state courts require for AI use in filings

The Oregon Judicial Department has not issued a statewide AI order. The Court of Appeals is addressing AI through proposed amendments to ORAP 1.40 and 13.25 (see Pending activity). At the appellate level, the Ringo and Doiban sanction tariff is already functioning as de facto court-imposed standards.

Has anyone been sanctioned for AI use in Oregon?

Ringo v. Colquhoun Design Studio, LLC, 345 Or App 301 (2025)

Oregon Court of Appeals, decided 2025-12-03. Counsel filed a brief with fabricated cases, a fabricated quotation, and fabricated substantive law. The court struck the brief. It found the fabrication “likely resulted from the use of artificial intelligence.” Sanctions totaled $2,000: $500 per fabricated citation and $1,000 per fabricated quotation or false statement of law. The court also imposed a forward-looking certification requirement. Any future brief from counsel must certify that it was drafted without generative AI, that counsel has read each cited case, and that every cited source exists.

Green Building Initiative, Inc. v. Green Globe Int’l, Inc., 350 F.R.D. 289 (D. Or. 2025)

U.S. District Court for the District of Oregon, Judge Michael Simon, October to November 2025. A large law firm cited one entirely fabricated case and one near-fabrication. The court issued a show cause order; the firm promptly acknowledged the errors and took voluntary remedial measures, and sanctions were not imposed. The case is the Oregon datapoint for the proposition that prompt disclosure and remediation can avert sanctions.

Couvrette v. Wisnovsky, 2025 WL 4109655 (D. Or. Dec. 12, 2025)

U.S. District Court for the District of Oregon, Medford Division, Magistrate Judge Mark D. Clarke. Liability and initial sanctions order 2025-12-12; final fees and fines order 2026-04-04. The court found 15 fabricated citations and 8 fabricated quotations across three briefs over five months. Counsel performed no Westlaw or Lexis verification and showed a “total lack of remorse.” Briefs were struck without leave to refile, claims were dismissed with prejudice, and the matter was referred to the Bar. Final monetary sanctions: $15,500 disciplinary fine plus $80,000 in attorney fees against lead counsel Stephen Brigandi, and $14,205.66 in attorney fees against Portland local counsel Tim Murphy. Murphy’s exposure arose from failure to meaningfully supervise out-of-jurisdiction counsel. Total approximately $110,000, the largest AI-related penalty imposed by an Oregon federal judge as of April 2026. Murphy’s separate liability shows that local counsel bears independent responsibility for verifying AI-generated content in filings they sign.

Doiban v. OLCC (Or Ct App, Mar. 18, 2026)

Oregon Court of Appeals, Presiding Judge Scott Shorr. Attorney Bill Ghiorso of Salem delegated brief-writing to staff. The paralegal used Google and Safari search engines, encountered AI-generated overviews that falsely confirmed fabricated cases as real, and copied them without Westlaw or Lexis verification. Brief filed November 2024 contained 15 fabricated citations and 9 fabricated quotations. Fabrications were identified April 2025. The sanctions hearing occurred November 2025. Sanctions: $10,000, a record for the Oregon Court of Appeals at the time. The full Ringo tariff would have reached $16,500, but it was capped because of recent medical difficulties. Key holdings: Google search is not adequate citation verification, because AI-generated overviews will confirm fabricated cases as real. An anti-AI policy is not a defense if actual verification was not performed. The supervising attorney is responsible for delegated research regardless of the tool used.

What AI-related rules are pending in Oregon?

As of April 16, 2026, the Oregon Court of Appeals ORAP Committee has proposed amendments to ORAP 1.40 and 13.25. Rather than regulating AI use directly, the proposal addresses the consequences of submitting fabricated law and facts. Sanctions triggers include nonexistent case citations, nonexistent quotations attributed to existing sources, unreasonably attributed statements of law, and factual assertions unreasonably attributed to the cited record. The proposal codifies the Ringo and Doiban tariff that is already functioning as de facto practice.

How do malpractice carriers in Oregon treat AI use?

Oregon is unusual nationally: malpractice coverage is mandatory for all private-practice attorneys whose principal office is in Oregon. Coverage is provided through the OSB Professional Liability Fund (PLF), so there is no “dominant private carrier” question for primary coverage. ORS 9.080(2)(a) authorizes the mandatory-coverage program, which operates under Article 18.2 of the OSB Bylaws. The PLF began operating July 1, 1978. PLF 2026 primary coverage is $300,000 per claim and $300,000 aggregate, plus a $75,000 claims expense allowance, with an annual assessment of $3,500 per licensee. PLF AI guidance directs lawyers to use only closed-system AI tools and flags open-model tools that retain prompts for training as higher risk. The Couvrette bar referral illustrates dual exposure: a sanctions order plus disciplinary action can affect future coverage eligibility, including any excess coverage placed above the PLF primary.

What does my Oregon malpractice carrier ask about AI at renewal?

The PLF, as the mandatory primary carrier, has not published an AI-specific application questionnaire. The substantive standard is Opinion 2025-205 plus the Ringo and Doiban tariff. At PLF assessment time or excess-coverage renewal, be ready to produce a written AI policy with closed-system default, citation verification logs through Westlaw or Lexis, vendor due diligence records, supervision and training records, and any incident file. Confirm specific underwriting items directly with the PLF and with any excess carrier.

The PLF has published two paired practice aids that operate as the de facto firm-policy framework for Oregon insureds. The first is the Generative AI Policy Development Guidelines for Law Firms (PDF, Rev. April 2026). The second is the companion Sample Generative AI Policy for Law Firms (DOCX, January 2026). Both are downloadable from the PLF Resources page (Services > CLEs & Resources > Practice Aids > Firm Operations > Office Systems and Procedures). The Sample Policy carries explicit permission for Oregon lawyers to use and modify the materials, but it is a starting framework rather than a complete operational document. PLF’s own caveat directs lawyers to tailor it to firm-specific operations.

What documentation should a Oregon firm keep on file?

Month one (foundational)

1. (Owner: managing partner + firm administrator) Written AI use policy designating closed-system tools as the default for client data. List prohibited uses, require Westlaw or Lexis verification before any AI-generated citation is filed, and define the supervision chain and incident procedure. Doiban shows that having the policy is not enough: verification must actually occur.
2. (Owner: litigation lead) Citation verification log for every filing using AI-generated research or citations, recording that each citation was verified in Westlaw, Lexis, or equivalent before submission. Paralegal-level Google search is not adequate (Doiban), and the Ringo certification language is a useful benchmark for appellate filings.
3. (Owner: managing partner) Supervision protocol for lawyers signing filings prepared by associates, paralegals, or out-of-jurisdiction co-counsel, since Couvrette and Doiban both turn on supervisory failure.

Months two and three (operational documentation)

1. (Owner: firm administrator + outside IT) Vendor due diligence record for each AI tool used with client information. Document whether the system is closed or open, confirm data retention and opt-out-of-training status, verify Oregon Consumer Privacy Act compliance, and note the date of review and responsible attorney.
2. (Owner: firm administrator) Attorney and staff training records showing review of Opinion 2025-205, the firm’s AI policy, and the Ringo and Doiban holdings. Required by Oregon RPCs 5.1 and 5.3.

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) Client confidentiality consent and AI disclosure in the engagement letter or addendum. Cover which tools may be used, what data may be submitted, and the associated risks. Obtain explicit consent for open-model tools and offer an alternative where the client objects.
2. (Owner: billing partner) Billing disclosures for AI tool costs in writing before any AI cost is charged. Per Oregon RPC 1.5 and Opinion 2025-205, hours must reflect actual time worked, not pre-AI baselines.
3. (Owner: matter lead attorney) Incident file documenting any discovered AI error, the remedy taken, and any required court or client notification. Green Building Initiative shows that prompt voluntary disclosure can prevent sanctions.

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