Washington State Needs AI Accountability Law, Tech Leader Argues

The accountability gap

When Washington's legislative session ended in March, lawmakers left unresolved a fundamental question: Who bears responsibility when an automated system denies someone a job, housing, or healthcare?

While existing statutes like the Washington Law Against Discrimination and Consumer Protection Act provide some coverage, the state lacks a dedicated framework for automated decision-making in high-stakes contexts. According to John deVadoss, co-founder of NeuralFabric Corp., this gap leaves residents vulnerable as AI systems increasingly shape consequential outcomes.

Automated tools already screen job candidates, evaluate creditworthiness, determine insurance eligibility, and influence decisions in healthcare, education, and government services. Yet Washington law cannot clearly answer basic questions about these systems: what technology was deployed, what role it played, whether it was assessed for discriminatory impact, and who holds legal responsibility for the outcome.

Why it matters

The absence of AI-specific accountability rules creates legal uncertainty precisely where the stakes are highest. As automated systems become standard infrastructure for employment, housing, and essential services, residents have no guaranteed path to understand or challenge decisions that affect their livelihoods. Federal preemption efforts add urgency—the Justice Department's AI Litigation Task Force has already intervened in Colorado's algorithmic discrimination case, signaling that state laws will face legal challenges.

What an AI Accountability Act should include

DeVadoss outlined a framework focused on genuinely consequential decisions in employment, housing, credit, insurance, healthcare, education, and government services. The proposal would establish a materiality threshold, applying only to systems that substantially influence outcomes while excluding incidental uses.

Core requirements would include predeployment impact assessments, plain-language notice to affected individuals, and meaningful human review rights. People subject to automated decisions would receive explanations in ordinary language: what system influenced the decision, what factors it considered, and how to challenge the outcome.

The framework would preserve private enforcement alongside state attorney general authority. DeVadoss argued that public enforcement alone cannot address individual harms—someone denied an apartment, medical authorization, or job opportunity needs a direct path to relief without waiting for statewide enforcement action.

Federal preemption threat

A December 2025 White House executive order directed federal agencies to challenge state AI laws and pursue uniform federal rules. While the order functions as an enforcement directive without inherent authority to displace state law, the legal threat has materialized. The Senate voted 99-1 to strip preemption language from a budget bill, and 36 state attorneys general opposed a federal moratorium on state AI laws.

DeVadoss recommended building legislation on traditional state authority over civil rights, consumer protection, and employment. The approach should focus on defined decision processes, factual disclosures, and documentation while avoiding content prescriptions. Including severability provisions would protect the statute if courts strike down individual provisions.

These details were first reported by the Seattle Times in an opinion piece by deVadoss.