When Washington lawmakers adjourned on March 12, they left the state without an answer to the question of AI accountability: Who is responsible when AI denies a person work, housing or other essential services?
Accountability gap
Washington’s general-purpose statutes cover some automated harms. The Washington Law Against Discrimination, the Consumer Protection Act and the My Health My Data Act, enacted for broader purposes, leave Washington without a dedicated framework for when an automated system shapes a high-stakes decision.
AI systems already operate in the settings where legal accountability matters most. Employers use automated tools to screen, rank and evaluate workers. Landlords and lenders rely on scoring systems that can affect access to housing and credit. Insurers, healthcare entities, schools and public agencies increasingly depend on automated systems. When AI dictates decisions, Washington law should be able to answer basic questions: What system was used, what role did it play, was it assessed for discriminatory impact, could the affected person seek human review and who is legally responsible for the result?
What the bill should address
The AI Accountability Act should begin with a disciplined scope. It should apply to genuinely consequential decisions: employment, housing, credit, insurance, healthcare, education and essential government services. It should include a materiality threshold that reaches automated systems that substantially influence an outcome, while excluding uses whose role is merely incidental. The statute should address public- and private-sector systems with equal precision.
The bill should require predeployment impact assessments, notice in plain language to anyone subject to an automated decision and a meaningful right to human review. Affected people should be told, in ordinary language, when an automated system influenced a decision, what type of system was used, what factors it considered and how the decision can be challenged.
Private enforcement should remain part of the framework and should be strengthened. The state attorney general should retain enforcement authority, but public enforcement alone cannot carry the full burden. A person denied a job, apartment, loan, medical authorization, insurance policy, school opportunity or public benefit should have a meaningful path to relief without waiting for the issue to become a statewide enforcement priority.
The bill should also fix a basic definitional problem. A high-stakes AI statute should reach beyond the consumer-protection frame. The statute should name people affected by these systems — job applicants, employees, tenants, patients, insureds — directly.
Federal preemption
A December 2025 executive order from the White House directs federal agencies and the Department of Justice to challenge state AI laws and to pursue a uniform federal framework. It should be understood as an executive enforcement directive with no capacity on its own to displace state law. The Senate stripped a preemption provision from the budget reconciliation bill, 99-1, and 36 state attorneys general have urged Congress not to revive a federal moratorium on state AI laws. The legal threat, however, has moved from abstraction to litigation. The Justice Department’s AI Litigation Task Force has been operational since January. The department’s April 24 intervention in xAI’s lawsuit against Colorado’s consumer-protection law aimed at preventing “algorithmic discrimination” shows that state AI laws will be tested.
A durable bill should rest on the state’s traditional authority over civil rights, consumer protection, employment and other government services. It should focus on defined high-stakes decision processes, factual disclosures, documentation and accountability, while avoiding efforts to prescribe model viewpoints or outputs. It should include severability, so that an adverse ruling on one provision leaves the rest of the act intact.
Washington state has long exercised authority to protect its residents from discrimination, unfair practices and unaccountable decision-making in matters of basic opportunity. AI makes the need for clear legal accountability more immediate.
