Washington: AI Ethics Guidance for Law Firms

Bottom line for a Washington firm: WSBA Advisory Opinion 202505 (Nov 20, 2025) maps all seven core professional duties to AI use and is the operative non-binding standard of care today. ALPS, the WSBA-endorsed malpractice carrier, has explicitly warned that standard LPL policies may not automatically cover AI-related claims, making documented oversight a coverage-protection issue. New legislation (HB 2225, effective January 1, 2027) creates consumer protection liability with a private right of action for firms using AI chatbots for client intake without adequate disclosure.

Start with: a written AI use policy, a citation verification protocol for court filings, and a vendor privacy review record for each AI tool in use.

WSBA Advisory Opinion 202505

Citation: WSBA Advisory Opinion 202505 (AO 2025-05), issued by the WSBA Committee on Professional Ethics. Dated November 20, 2025 (PDF metadata); publicly highlighted January 20, 2026. Primary source PDF. Companion article. Advisory opinions database.

Status: Advisory opinion. Per WSBA policy, advisory opinions reflect the CPE’s position but are not individually approved by the Board and do not reflect the official WSBA position. Not binding rules.

The opinion defines AI broadly to include generative AI, agentic AI (functioning as an agent for the user), and autonomous AI (operating with substantial independence). It deliberately avoids product-specific guidance.

What it requires

• Lawyers must understand AI tools they use, including capabilities, limitations, and risks. All AI-generated work remains the lawyer’s professional responsibility (RPC 1.1, Comment 8).
• Thorough review and verification of AI-generated outputs is required. Overreliance on AI without proper oversight is an ethical violation (RPC 1.3).
• AI tools, including consumer applications, must have adequate safeguards preventing unauthorized disclosure of client data. Vendor terms of service must be reviewed before use (RPC 1.6).
• Full responsibility for accuracy in court filings is required even when AI drafted content. AI hallucinations resulting in fabricated citations are specifically addressed (RPC 3.3).
• Supervising attorneys and firm managers must ensure subordinate lawyers and nonlawyer staff receive adequate training in ethical AI use. Inadequate training of a supervised person who causes harm is itself an ethical violation (RPCs 5.1, 5.3).
• Lawyers cannot bill for “time saved” by AI. Overhead costs cannot be passed along to clients without consent. The basis of fees and expenses must be explained at the outset of representation (RPC 1.5(a), (b)).

What it recommends

• Consult clients about AI use, particularly when sensitive client information is involved; informed consent may be required and documentation in engagement agreements is recommended (RPC 1.4).

Rules cited

RPC 1.1 (including Comment 8); RPC 1.3; RPC 1.4; RPC 1.5(a) and (b); RPC 1.6; RPC 3.3; RPC 5.1; RPC 5.3.

Notable gaps

Several issues fall outside the opinion’s scope. Whether AI use must be disclosed to courts absent a specific court order is not addressed. AI-generated evidence and authentication issues are not addressed. Agentic AI autonomously communicating with clients under the lawyer’s identity receives no guidance. The line between situations requiring the client’s informed consent and those where consultation suffices under RPC 1.4 is not precisely drawn.

WSBA Legal Technology Task Force Final Report

In September 2025, the Board of Governors accepted the WSBA Legal Technology Task Force Final Report and Recommendations; it was updated February 5, 2026. The accompanying member survey ran in October 2024 with 516 respondents. It found only 25% of Washington legal professionals use generative AI regularly, only 22% adoption among small-firm lawyers, and only 23% of members believe current ethical rules adequately cover generative AI use. The report recommended building AI literacy, creating a WSBA Technology Resource Hub, and establishing cybersecurity standards affordable to small practices. A proposed Washington Supreme Court-affiliated oversight board has not yet been established.

What federal courts in Washington require for AI use in filings

No district-wide AI standing order has been adopted in the Western District of Washington (WAWD) or the Eastern District. Individual judges may impose disclosure requirements in chambers practices. Firms should review the assigned judge’s standing orders at the opening of each federal matter. The operative federal default is Fed. R. Civ. P. 11.

What Washington state courts require for AI use in filings

No Washington Supreme Court administrative order on AI in court filings has been identified. At the local level, Clallam County District Court II Local Rule LARLJ 49 is captioned “Use of Artificial Intelligence in Court Submissions.” It requires attorneys and parties to disclose any use of AI in the creation or editing of documents or evidence submitted to that court. Disclosure must be made at the time of submission via a certification attached to the document, identifying the AI technology used and certifying the attorney’s final review and approval. Violations may be sanctioned, including under CrRLJ 4.7(g)(7). The rule does not displace the attorney’s duties of competence, diligence, and supervision under the Washington RPCs. This is a local rule for District Court II only. Confirm the assigned court’s standing orders before filing elsewhere.

LeDoux v. Outliers, Inc. (W.D. Wash. 2026)

Citation: LeDoux v. Outliers, Inc., No. 3:24-cv-05808-TMC (W.D. Wash. July 24, 2026) (Cartwright, J.) (Order on Sanctions, Dkt. 265); see id. (Feb. 4, 2026) (Order to Show Cause, Dkt. 209).

In February 2026 the court found that the plaintiff had submitted “dozens of inaccurate factual and legal citations” across multiple filings and ordered her attorney, Jocelyn Stewart, to show cause by sworn declaration why she should not be sanctioned under Fed. R. Civ. P. 11. Stewart’s declaration stated that she had used ChatGPT and Claude. On July 24, 2026 the court found her conduct “tantamount to bad faith,” personally sanctioned her $3,000, required her to certify in all later briefing in the case that all citations have been verified, directed her to give the order to her client, and discharged the show-cause order. On August 18, 2026 (Dkt. 269), the court excluded one of the plaintiff’s experts, finding that “the hallucinated citations undermine Dr. Holguin’s entire report,” granted summary judgment, and dismissed the remaining claims with prejudice. See the tracker entry. No W.D. Wash. district-wide AI standing order resulted from this case. Rule 11 sanctions are generally not covered by LPL policies. That sharpens the coverage exposure for verification failures.

What AI-related rules are pending in Washington?

Proposed AI comments to the RPC. At its July 24-25, 2026 meeting, the WSBA Board of Governors approved a proposal by the Committee on Professional Ethics that the WSBA serve as co-proponent with the Washington Supreme Court Clerk in submitting amendments to the commentary to RPC 1, 3, 5, 7, and 8 on the legal ethics of AI (July 2026 Board recap). According to the recap, the Court first developed the drafts, and the Clerk and the Committee then refined them; the comments “cover use, disclosure, firm policymaking and supervision, fees, bias, and accurate review of AI in providing legal services.” The recap describes the comments’ scope but does not quote their text. As of September 28, 2026, the Supreme Court’s proposed rules page does not list them for public comment. Nothing binds Washington lawyers until the Court publishes the comments, takes comment, and adopts them.

Governor Ferguson signed HB 2225 (AI Companion Chatbot Regulation Act) on March 24, 2026; it takes effect January 1, 2027. The act requires disclosure that any chatbot is artificial and creates a private right of action under the Washington Consumer Protection Act. For law firms using AI-powered client intake chatbots without adequate disclosure, the direct risk is consumer protection liability and potential UPL exposure.

HB 1170 (AI Content Transparency Act) is enacted and effective February 1, 2027. AI operators must inform users when content is developed or modified by AI, applying to AI tools used in client-facing communications.

The Washington AG AI Task Force, established by ESSB 5838, released its final report on July 1, 2026. Across its three reports the Task Force made 11 policy recommendations to the Legislature and Governor, which may inform future legislation affecting legal practice.

How do malpractice carriers in Washington treat AI use?

ALPS (Attorneys Liability Protection Society) is the WSBA-endorsed malpractice carrier. In its Insurance Coverage Issues for Lawyers in the Era of Generative AI, ALPS warns that standard LPL policies do not currently have explicit AI exclusions. AI-related claims face two existing coverage challenges. The first is that “no professional services were rendered” if a lawyer allows AI to make legal judgments without attorney review. The second is unauthorized practice exclusion when AI makes independent legal determinations without attorney oversight. Confidentiality breaches from feeding client data into public AI tools may implicate intentional-act exclusions. ALPS directs firms to maintain attorney review and validation of all AI outputs, use only AI platforms with strict data privacy terms, and document oversight and review processes.

Washington RPC 1.4(c) requires written client disclosure if a lawyer does not carry malpractice insurance meeting the $100,000 per occurrence / $300,000 aggregate threshold, with written informed client consent acknowledging the gap. If coverage lapses during representation, written disclosure to all current clients is required within 30 days.

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

ALPS has not published a specific AI application questionnaire in materials reviewed for this entry. AO 202505 establishes the substantive standard, and the ALPS coverage-issues guidance points to the same conduct: attorney review, vendor diligence, and documented oversight. Plan on producing five artifacts. A written AI use policy. Vendor privacy review records. Citation verification logs. Training records. Billing documentation that excludes “time saved” by AI. At each renewal cycle, document an ALPS coverage review confirming whether the current policy addresses AI-related errors. Confirm application items directly with ALPS or your broker.

What documentation should a Washington firm keep on file?

Month one (foundational)

1. (Owner: managing partner + firm administrator) Written AI use policy covering approved platforms, prohibited uses (inputting client-identifying information into public tools), review-before-filing requirements, and supervision protocol. Maps to RPC 1.1 and RPCs 5.1, 5.3 as required by AO 202505.
2. (Owner: litigation lead) Citation verification protocol documenting that all AI-generated citations are independently verified before any court filing. Directly responsive to LeDoux v. Outliers (W.D. Wash. 2026), Fed. R. Civ. P. 11, and RPC 3.3.
3. (Owner: firm administrator + outside IT) Vendor privacy review documentation for each AI platform: terms of service review, data retention and training data use, opt-out options, security certifications. Maps to RPC 1.6 and the ALPS coverage guidance.

Months two and three (operational documentation)

1. (Owner: firm administrator) Staff training records for all attorneys and staff using AI tools, including hallucination risks, RPC 3.3 candor obligations, and the firm’s AI policy. Maps to RPCs 5.1 and 5.3; AO 202505 treats inadequate training of a supervised person as itself an ethical violation.
2. (Owner: managing partner + billing partner) Engagement letter AI disclosure language disclosing potential AI use, firm confidentiality safeguards, and client consent (or right to withhold consent). Maps to RPC 1.4.
3. (Owner: billing partner) Billing documentation showing AI use is accounted for using actual attorney review time only, with no AI processing time billed and no overhead cost pass-through without client consent. Maps to RPC 1.5; AO 202505 explicitly prohibits billing for “time saved” by AI.

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 before client-identifying information enters a third-party AI platform; document tool, information shared, and purpose. Maps to RPC 1.6 and RPC 1.4.
2. (Owner: managing partner) ALPS coverage review at each renewal cycle. Document whether the current policy addresses AI-related errors, given ALPS’s published warning that standard policies may not automatically cover AI-related claims.
3. (Owner: managing partner + firm administrator) HB 2225 intake chatbot review before January 1, 2027. Confirm any client-facing AI chatbot discloses its AI nature, or shut it down. Private right of action under the Washington Consumer Protection Act.

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