Washington State DOL's AI Phone System Reportedly Failed to Provide Spanish-Language Service to Callers Requesting Spanish

For months, callers to the Washington State Department of Licensing who selected Spanish reportedly received AI-generated English responses spoken with a Spanish accent rather than actual Spanish-language service. The agency reportedly apologized and said staff configuration caused the error, which purportedly created accessibility problems for callers seeking language support.

Risk classification

• Primary risk domain: 1 Discrimination & Toxicity
• Primary risk subdomain: 1.3 Unequal performance across groups

The AI-driven phone system failed to perform effectively for Spanish-speaking users compared to English-speaking users, resulting in unequal outcomes and reduced benefits based on language group membership.

Additional risk subdomains

• 7.3 Lack of capability or robustness: The system exhibited a technical failure where it outputted English with a Spanish accent instead of translating the text into Spanish.

Causal factors

• Entity: Human
• Intent: Unintentional
• Timing: Post-deployment

The incident was caused by human configuration errors made by DOL staff when setting up the deployed AI-driven phone system.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The system is an AI-driven automated voice/chatbot service, which falls under transparency obligations where users must be informed they are interacting with an AI.

AI system and alleged parties

• AI system: Amazon Web Services (Amazon)
• AI purpose: Voice Generation; AI Voice Assistant
• Behaviour type: Tool
• Alleged developer: Amazon
• Alleged deployer: Washington State Department of Licensing, Amazon
• Alleged harmed parties: Spanish language speakers, General public of Washington State, General public

Harm severity

Highest direct severity in any category: Substantial. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Minor, indirect Minor
• Civil rights: direct Minor, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Differential treatment

Reported: The report explicitly describes differential treatment of Spanish-speaking individuals caused directly by the incident.

Directly caused: Spanish-speaking callers received English with a Spanish accent instead of Spanish translation, while English speakers received proper service.

Indirectly caused: Non-English speakers had to wait longer or find workarounds to access licensing services.

Inferred additional harm: Thousands of limited-English Spanish speakers in Washington State likely faced exclusion or significant delays in obtaining driver's licenses or other DOL services.

Civil rights

Reported: The report explicitly describes accessibility issues for people who call in every day and need to speak in a different language.

Directly caused: Temporary denial of equal access to government services for Spanish speakers.

Indirectly caused: N/A

Inferred additional harm: Potential systemic violation of state or federal language access requirements for government agencies.

People affected

• Occurrences reported: 1
• People reportedly harmed: 2
• People reportedly exposed: 2

Potential causes

Management

• Lack of Project Oversight: Management deployed the system without ensuring proper quality checks.
• Poor Feedback Monitoring: No mechanism was active to capture and act on early user complaints.

Technology

• Newer AI Platform Integration: The self-service option runs on newer, AI-driven technology.
• Incorrect Voice Model Selection: An AWS voice mimicking Castilian Spanish was selected for English text.

Data Inputs

• Untranslated English Text: The system provided English text instead of translating it to Spanish.
• Incomplete Translation Variables: Only numbers like 'tres' were translated while main text remained English.

Human Factors

• DOL Staff Configuration Error: Agency staff misconfigured the settings on the AWS phone platform.
• Lack of User Verification: Staff failed to check how the voice sounded to Spanish-speaking callers.

Process and Methods

• Inadequate Pre-testing: Testing on all language options was not completed before deployment.
• Delayed Issue Resolution: The system glitch persisted for months without being identified or fixed.

Information quality

• Classification confidence: High
• Reason for confidence: The reports are consistent, from reputable sources, and clearly describe the technical glitch, its impact, and the agency's response. The technical cause is identified as human configuration error.
• Ambiguities identified: The exact technical configuration error within AWS is not fully detailed, but the general cause is clear.

An administrative and technical configuration error in the Washington State Department of Licensing phone system resulted in accessibility barriers for Spanish-speaking residents. While it caused localized public service delivery issues, it has negligible national security implications and is being managed through standard administrative procedures.

• Overall national security impact: Minor
• Response level: Moderate
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. The issue has been temporarily disabled and is being resolved administratively, presenting no ongoing or immediate crisis.
• Autonomy: Human-controlled. The system is a text-to-speech tool configured by human staff, acting without independent decision-making capabilities.
• Novelty: Established threat. Configuration errors and software translation bugs in automated systems are well-established issues.

Impact by dimension

• Physical security: Negligible. No threats to physical systems, critical infrastructure, or human safety were indicated in the incident reports.
• Information security: Negligible. No information warfare, intelligence compromise, or systematic information manipulation occurred.
• Sovereignty: Minor. Minor localized disruption to state government service delivery causing accessibility barriers, managed through standard administrative adjustments.
• Economic security: Negligible. No impact on strategic industries, economic stability, or technological competitive advantage.
• Societal stability: Minor. Localized accessibility barriers and unequal service performance for Spanish-speaking residents, but no threat to large-scale societal stability or national security.