Senate Bill 6254 would amend RCW 34.05.010 and RCW 43.42.030 and add new sections to chapters 43.42 and 34.05 RCW to require state agencies and the Office of Regulatory Assistance to use artificial intelligence and human review to analyze and reduce regulatory burdens. The bill adds or clarifies many definitions (including “agency,” “artificial intelligence,” and “machine learning”) and creates new procedural obligations: the office must obtain a third‑party, AI‑generated regulatory reduction report by December 31, 2026; each agency (as defined in the bill to include state boards, commissions, departments, institutions of higher education, and similar officers authorized to make rules or conduct adjudicative proceedings, excluding the legislative and judicial branches and certain executive offices) must report intended actions within four weeks of receiving the report, begin report‑supported regulatory actions and provide a future review schedule within four weeks of that notice, and must seek to leverage AI by December 31, 2026 and every four years thereafter to analyze rules for whether requirements are mandated by binding law, redundant or in conflict, comparable to neighboring states, and where text can be streamlined. The bill also requires agencies, after final legal review, to eliminate rules or requirements that conflict with binding law or duplicate existing law and to use the expedited rule‑making process in RCW 34.05.353 when possible, with those actions to be completed no later than June 30, 2027; and it requires agencies to establish by December 31, 2026 a schedule to review all guidance documents at least once every four years, analyzing guidance for legal consistency and removing provisions that create new binding legal requirements within six months after each review. These are procedural and administrative changes imposing mandatory review and revision duties on agencies rather than creating new criminal penalties. The text as provided leaves some implementation details unspecified: the bill calls for a “third‑party” AI analysis but does not say how third parties will be selected or what standards or methods the AI must meet; it repeatedly requires a “final legal review” but does not identify who performs that review; and a definition that begins for “permit agency” is cut off. The bill was read for the first time on January 21, 2026 and referred to the Committee on State Government, Tribal Affairs & Elections.

If enacted, state agencies and the office of regulatory assistance will have to use outside AI analysis and their own AI-assisted reviews to hunt for rules and guidance that conflict with law, duplicate other requirements, or can be shortened, with a statewide AI report due December 31, 2026 and agency rule-and-guidance review schedules set the same day and repeated every four years. Agencies must tell the office within four weeks how they plan to act on the report, begin the recommended changes soon after, and complete legal cleanups of conflicting or duplicative rules by mid‑2027 and fix guidance documents within six months after each review cycle. This will most directly affect agencies that write rules (including state departments and public higher education institutions) and the office of regulatory assistance, and it gives agencies the option to use an expedited rule‑making process where appropriate. Practically, agencies will likely need to hire or contract with third‑party AI vendors, assign staff for legal review and project management, and absorb short‑term costs and workload spikes to meet tight deadlines and ongoing four‑year cycles; the office must likewise manage the initial third‑party report. The requirement could reduce regulatory burdens over time by removing conflicts and redundancy, but the bill leaves key details open—who selects and oversees the third‑party AI, what standards the AI must meet, and exactly who performs the “final legal review”—which creates implementation and legal‑risk uncertainty for agencies planning budgets and timelines.