HB 2667
In CommitteeHouse
AI consumer protections
Providing consumer protections for artificial intelligence systems.
This status may be delayed. See Action History below for the latest updates.
How does a bill become law?
- Introduced: The bill is filed and assigned a number.
- Committee: A subject-matter committee holds hearings, takes public testimony, and decides whether to advance the bill.
- Floor Vote: The full chamber (House or Senate) debates and votes on the bill.
- Opposite Chamber: The bill repeats the committee and floor vote process in the other chamber.
- Governor: The Governor reviews the bill and decides whether to sign or veto it.
- Signed: The bill has been signed into law.
AI Analysis
This bill creates a comprehensive framework to protect Washington consumers from harmful or discriminatory uses of artificial intelligence, especially in high-stakes decisions like employment, housing, and credit. It requires businesses and government agencies to disclose AI use, assess risks, and implement safeguards, while preserving space for innovation.
- Requires businesses that deploy high-risk AI systems (e.g., for employment, housing, or lending decisions) to implement risk management policies and conduct impact assessments to prevent algorithmic discrimination.
- Mandates that deployers notify consumers before using AI to make consequential decisions, and provide clear explanations of the AI’s purpose and contact information.
- Establishes a rebuttable presumption that deployers who comply with the law used reasonable care to prevent discrimination.
- Requires government agencies to disclose when consumers are interacting with AI, using clear, non-deceptive language.
- Extends and expands the existing AI Task Force to study AI in the workplace and deliver recommendations to the legislature by July 2027.
- Exempts small businesses (fewer than 50 full-time employees) from certain requirements if they use third-party AI without training it on their own data.
Who is affected
- Businesses and organizations that deploy high-risk AI systems — Businesses that use high-risk AI systems to make consequential decisions (e.g., hiring, housing, lending, or government services) must implement risk management policies, conduct impact assessments, and notify consumers about AI use.
- Washington consumers — Consumers in Washington will receive clear notice before AI is used to make decisions affecting their access to jobs, housing, education, credit, or government services, and have the right to know how the AI works.
- Government agencies — State and local government agencies must disclose when AI is used in consumer-facing interactions, and must follow transparency and accountability requirements for AI systems.
- AI developers and technology companies — AI developers and technology companies must ensure their high-risk AI systems are designed to reduce bias and discrimination, and may be required to share impact assessments with deployers under certain conditions.
- Workers and labor advocates — Workers and labor advocates will benefit from a new advisory group focused on ensuring AI in the workplace is fair, transparent, and protective of employee rights.
Pro/Con Analysis
Stronger case for benefits
Potential Benefits (5)
Mandatory consumer notice and plain-language explanation before consequential AI decisions (e.g., job screening, loan denial, housing application) empowers individuals to challenge unfair outcomes, request human review, and exercise meaningful consent — a foundational right in algorithmic decision-making that was previously absent.
Rights & LibertiesPeopleRef: Sec. 6 & Sec. 7The requirement to conduct impact assessments and implement risk management policies specifically targeting algorithmic discrimination creates a structural safeguard against biased outcomes in employment, housing, and credit — directly protecting historically marginalized groups from automated bias that often replicates or amplifies human prejudice.
Rights & LibertiesPeopleRef: Sec. 3(1)(a) & Sec. 5(2)(b)The task force’s requirement to include advocates representing communities disproportionately harmed by algorithmic bias — including LGBTQ+, racial/ethnic minorities, people with disabilities, and veterans — ensures that regulatory design and implementation are informed by lived experience, not just technical expertise.
Public SafetyPeopleRef: Sec. 11(2)(c)(viii), (ix), (xiii)The new AI Workplace Advisory Group, with mandatory labor and employee representation, will produce concrete recommendations on mitigating workplace AI harms — directly benefiting workers by embedding labor voices in the development of AI governance for employment tools.
Business & EmploymentPeopleRef: Sec. 14 & Sec. 11(4)(a)Requiring government agencies to disclose AI use in consumer-facing interactions (e.g., automated benefits determinations, chatbots) increases transparency and accountability in public services — helping residents understand when decisions affecting their lives are made by machines, not humans.
Local GovernmentPeopleRef: Sec. 10
Potential Concerns (5)
The exemption for small businesses (under 50 FTEs) that use third-party AI without training it on their own data significantly reduces compliance burdens for micro-businesses and sole proprietors, but the threshold is high enough that many small employers and service providers still fall under the full requirements, and the exemption does not apply if they modify or train the AI themselves — limiting its practical utility for many small operators.
Business & EmploymentIndustryRef: Sec. 5(6)The bill creates new legal exposure for deployers by making violations of its provisions actionable under the Consumer Protection Act (CPA), which allows for statutory penalties, attorney fees, and class-like enforcement — a risk disproportionately borne by small and mid-sized businesses lacking legal counsel, while large firms can absorb compliance costs more easily.
Business & EmploymentIndustryRef: Sec. 9(1)(a) & Sec. 7The repeated carve-outs protecting trade secrets and proprietary information do not eliminate the requirement to produce internal risk assessments, impact analyses, and audit trails — potentially exposing competitive strategies to competitors or litigants via discovery in enforcement actions, even if the raw model weights or training data remain protected.
Business & EmploymentIndustryRef: Sec. 5(7) & Sec. 4(3)Mandating clear, non-deceptive AI disclosures for all government-consumer interactions (e.g., automated benefits eligibility, chatbots for DMV services) will require new staff time, system redesign, and ongoing compliance monitoring — costs that strain already-constrained local governments and public service agencies.
Local GovernmentLean industryRef: Sec. 10The bill explicitly preserves existing obligations under data privacy and security laws (e.g., CCPA, HIPAA, WPA), but does not harmonize with them — creating overlapping, potentially conflicting compliance obligations for entities operating across sectors (e.g., health insurers using AI for underwriting), increasing legal complexity and cost.
Business & EmploymentIndustryRef: Sec. 9(2)
Who Is Most Affected
Workers and job applicants benefit significantly — especially those in vulnerable positions (e.g., gig workers, low-wage earners, people of color) — as the bill mandates transparency and bias mitigation in AI-driven hiring, performance evaluation, and termination tools. However, small employers using third-party AI screening tools may face higher costs to remain compliant.
Consumers in housing, credit, and insurance markets gain strong protections against algorithmic discrimination and receive clear notice before AI affects their access to essential services. However, small landlords, lenders, and insurers may face new compliance costs and liability exposure, potentially reducing service availability in rural or underserved areas.
State and local governments gain a framework to ensure transparency and fairness in automated public services (e.g., benefits, licensing, law enforcement risk tools), but must allocate staff time and IT resources to implement disclosure and audit protocols — straining already tight budgets, especially in smaller jurisdictions.
Large tech firms and AI developers may benefit from regulatory certainty and a competitive advantage over smaller competitors who lack compliance resources. However, they face new obligations to support deployers with impact assessments and may face increased litigation risk under CPA enforcement. The bill’s broad definitions could also restrict certain open-source or research uses.
Small businesses under 50 FTEs using third-party AI without training it themselves are exempt from core requirements — a clear win. But those just above the threshold (e.g., 50–100 FTEs) or using custom AI models face steep compliance costs (assessments, policies, monitoring), potentially reducing competitiveness and innovation capacity.