Regulates City departments' acquisition and use of generative AI, requiring Seattle IT to approve exceptions and conduct risk reviews. Mandates human review of AI outputs, attribution, and bias reduction. Prohibits AI training using City data without controls.
Analysis summaries, actor details, and coverage mappings were LLM-classified and may contain errors.
This is an internal municipal government policy document that establishes binding requirements for City of Seattle departments and employees regarding generative AI use. While it has mandatory language and enforcement mechanisms for internal compliance, it is not a statute or regulation with external legal force.
The document has good coverage of approximately 8-10 subdomains, with strong focus on discrimination and bias (1.1, 1.3), privacy compromise (2.1), false information (3.1), security vulnerabilities (2.2), overreliance (5.1), lack of transparency (7.4), and lack of robustness (7.3). Coverage is concentrated in fairness, privacy, misinformation prevention, and AI system reliability domains.
This is an internal municipal government policy that governs AI use within the City of Seattle's own operations. The primary sector governed is Public Administration (excluding National Security), as the policy applies to all City departments, employees, vendors, and contractors operating on behalf of the City government. The policy does not regulate external private sector entities or other industries.
The document primarily focuses on the acquisition, deployment, and operational use stages of generative AI systems. It establishes requirements for technology acquisition approval, mandates human review of AI outputs during operation, and requires ongoing monitoring through documentation and records retention. The policy does not substantially address data collection, model building, or pre-deployment verification stages.
The document explicitly defines and focuses on 'generative artificial intelligence' as its primary scope, describing it as systems using large language models capable of generating new content including text, images, video, and audio. It also references AI technologies generally through its principles section but applies policy statements only to generative AI systems. No mention of frontier AI, general purpose AI, foundation models, compute thresholds, or open-weight models.
City of Seattle, Data Privacy, Accountability and Compliance (DPAC) division, Director of DPAC, City of Seattle Chief Privacy Officer
The policy document identifies DPAC division as responsible for maintaining the policy, with ownership by the Director of DPAC and Chief Privacy Officer. The document is issued by the City of Seattle government.
Chief Technology Officer (CTO), Seattle IT, individual division directors, department leaders
The CTO is explicitly designated as responsible for compliance enforcement. Seattle IT has approval authority over generative AI acquisitions and can revoke authorizations. Division directors and department leaders have authority to impose disciplinary actions.
Seattle IT, Data Privacy, Accountability and Compliance (DPAC) division, departmental RSJI Change Teams
Seattle IT conducts risk and impact reviews of generative AI technology requests. DPAC maintains the policy and creates risk criteria. Departments are required to document HITL reviews and maintain records, with RSJI Change Teams involved in equity evaluations.
All City of Seattle departments, City employees, vendors, contractors, and volunteers who operate on behalf of the City
The policy explicitly states its scope covers all City departments and extends to vendors, contractors, and volunteers operating on behalf of the City. City employees are identified as users of generative AI tools who must comply with the policy requirements.
10 subdomains (7 Good, 3 Minimal)