AI Guiding Principles
- Human-Centric: We will leverage artificial intelligence to enhance the delivery of public services, prioritizing user needs, accessibility and responsiveness. Our AI systems will be equitable and capable of consistently providing high-quality, accurate and reliable outcomes for all community members.
- Transparency and Accountability: Bellevue believes public confidence in AI depends on transparency and accountability. We will provide transparent AI decision-making processes, clearly communicating how AI systems operate and how decisions are made.
- Inclusion and Non-Discrimination: We will implement AI systems that promote inclusivity and equitable access to city services, information and opportunities. We will evaluate AI systems for bias and address discriminatory impacts.
- Innovation: Bellevue encourages creativity, continuous learning and responsible experimentation that address community needs. Our AI technologies will aim to improve city services, operations and problem-solving in transformative ways.
- Stewardship: Recognizing the environmental impacts of AI, we will evaluate AI systems through the lens of the city’s environmental and resource management goals. We will prioritize AI solutions that reduce waste, enhance long-term sustainability and optimize operations.
- Privacy and Security: Bellevue safeguards the security and privacy of data as well as the city’s infrastructure to maintain trust and stability. We will reduce privacy and security risks to the greatest extent possible while ensuring compliance with all applicable laws and regulations.
- Ethical Use: Our AI technologies will uphold the city’s core values and comply with all applicable laws, regulations and ethical AI use.
Artificial Intelligence (AI) Policy
Purpose
The purpose of the City of Bellevue's artificial intelligence policy is to establish clear guidelines and governance for the ethical, transparent and responsible use of AI technologies, including generative AI systems. This policy aims to facilitate innovation and improve the delivery of public services while safeguarding privacy, ensuring compliance with applicable laws, protecting against bias and discrimination, and maintaining public trust.
Scope
This policy applies to:
- All AI systems, including generative AI tools (such as ChatGPT, Copilot or other public or proprietary applications), deployed or utilized by the City of Bellevue or on its behalf.
- All individuals and entities, including—employees, officials, consultants, contractors, partners, vendors, volunteers or third parties—who access, develop, deploy, operate, maintain or interact with AI systems and generative AI applications as part of their role or responsibilities related to the City of Bellevue.
- Use of AI on all devices, including city-issued and personal devices, when used to conduct city-related business, services or operations.
- All data and information processed by, input into or derived from AI systems, especially data involving sensitive, proprietary or personal information.
Definitions
Artificial Intelligence
Artificial intelligence is a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments. AI systems use machine- and human-based inputs to perceive real and virtual environments; abstract such perceptions into models through analysis in an automated manner; and use model inference to formulate options for information or action.
AI system
Any system, software, sensor or process that automatically generates outputs including, but not limited to, predictions, recommendations or decisions that augment or replace human decision-making. This extends to software, hardware, algorithms and data generated by these systems, used to automate large-scale processes or analyze large data sets.
Confidential or Sensitive Data
Information protected by law or designated for restricted official use, requiring careful handling to prevent unauthorized disclosure. This includes legally protected personal or employee data and data with specific statutory, regulatory or contractual protections, as well as internally designated sensitive materials intended for limited use.
Generative AI (GenAI)
A subset of artificial intelligence capable of synthesizing new, original content— such as text, images, audio or other forms of media—in response to user prompts. GenAI models, like ChatGPT or Copilot, can function as standalone tools or can be embedded into other software applications.
Responsive generative AI tools (Responsive)
AI tools that interact directly with community members and employees to provide information from the city’s website or similar discreet set of city-generated or city-curated content.
Supportive generative AI tools (Supportive)
AI tools capable of generating text, images or other data from sources not limited to city-generated content in response to employee-provided prompts.
Policy Requirements
Procurement of AI Technology
- All procurement of AI technology must follow the city's AI procurement process and related policies (e.g., New Technology Request Process).
- If a technology that has already been approved for use adds or incorporates AI capabilities, additional review and approval may be required to use those capabilities.
- All vendors and third parties who develop, deploy or operate AI systems on behalf of the city will comply with city-established AI guidelines, privacy standards and contractual obligations, including completion of required documentation and agreements.
Use of AI and AI Outputs
- Employees are prohibited from attempting to gain access to or use unapproved AI applications when using the city's systems or networks, conducting business on behalf of the city, or when using the city's data. The city will maintain a list of approved AI applications that will be accessible to employees.
- The city encourages pilot opportunities that allows for testing and experimenting with new AI technologies, while assessing risks in a controlled environment, to develop comprehensive AI solutions.
- All city use of AI systems must be consistent with this policy, the AI Guiding Principles, and in accordance with all applicable local, state and federal laws and existing City of Bellevue policies, including the Technology Resource Usage Policy and Work Rules (TRUP).
- Supportive AI-generated work and outputs must be reviewed prior to use in public-facing content. Users should verify that the work product does not contain inaccurate or outdated information, potentially harmful or offensive material, and is compliant with all city policies and applicable laws.
- Responsive AI-generated work and outputs should be randomly audited by a department-designated employee to help ensure accuracy and appropriateness of the communication(s) on a regular schedule as determined by the relevant department.
Transparency
- Supportive AI-generated content (including images, video, text or other data) that is used substantively in a final public-facing work product must be attributed to the relevant AI system and where possible, identify the department or group who reviewed the content prior to publication. Example: Some material in this brochure was generated using ChatGPT 4.0 and was reviewed and edited by a member of the Parks & Community Services Department before publication.
- Responsive AI-generated content that cannot be reviewed prior to use (e.g., outputs from an AI chatbot that assists the public in searching the city’s website to answer questions), must include a statement notifying the public that they are interacting with an AI system or identifying the AI system used, that the content was not reviewed by a human, and contact information the public can use if they have questions or want more information. Example: You are interacting with an AI chatbot created by the City of Bellevue to help the public better find content on our website. The city has not reviewed the content generated by the chatbot and cannot guarantee it is accurate. Please contact [email or phone number] if you need more information.
- The City will maintain detailed technical documentation for all AI systems, including vendor provided or internally developed Responsible AI Risk Assessments.
Reducing Bias and Harm
- Use of AI tools must be consistent with the city's core value of diversity, equity, and inclusion, and departments are encouraged to work with their diversity liaison team members when considering using AI tools that will inform decisions or policy to discuss potential impacts.
- If it appears that AI systems are producing biased outputs or that use of AI tools are having discriminatory impacts, these concerns must be immediately escalated to your designated diversity liaison to review and determine whether further review by the Cyber Incident Response Team is necessary.
Data Privacy and Security
- Confidential and sensitive information may not be integrated, entered or otherwise incorporated into any AI software or service without an approved policy exception.
- There is no right to privacy in the use of city technology resources, and the city reserves the right to access and monitor the use of GenAI applications on any city-issued devices or that appears on company-managed networks to ensure compliant use of these systems.
Periodic Reviews of AI Systems
- The city will conduct structured AI reviews, including risk assessments, prior to deployment and periodically thereafter, to identify and mitigate potential ethical, operational and compliance risks associated with AI systems.
- The reviews will be conducted by the IT Security Team, Diversity Liaison Team, Public Records Team and others as necessary based on risk assessments.
- The scope will include a review of data privacy, security measures, bias and discrimination checks and compliance with applicable laws and city policies.
- Periodic reviews may include a report outlining recommendations, potential corrective actions and mitigations, or policy updates if needed.
Training and Education
- Prior to using AI systems, to ensure responsible and effective use, City of Bellevue employees will complete training that will provide an overview of AI technologies, including generative AI systems, their capabilities and their potential impact on public services.
- AI training will emphasize the ethical considerations and responsibilities associated with using AI, including safeguarding privacy, ensuring compliance with applicable laws and protecting against bias and discrimination.
Community Engagement and Awareness
- Bellevue will explore partnerships with businesses, nonprofit organizations and other governmental entities on the development of innovative solutions to enhance or improve delivery of city services, support digital inclusion opportunities, assess emerging innovations and technologies, and advance the city’s economic goals by cultivating and supporting the startup ecosystem.
- The city will provide information and resources to the public about the AI technologies being used, their purposes and their potential impact on public services. Additionally, the city will establish feedback channels for community members to provide input. This will be leveraged to improve AI policies and practices within the city.
Public Records and Records Management
- All records prepared, owned, used or retained by the city in AI technology for city business are public records and must be maintained consistent with the city's record retention schedule and disclosed upon request.
- All AI solutions and/or vendors approved for city use will be required to support retrieval and export of public records upon request.
Administration, Reporting and Violations
- As provided in the TRUP, inappropriate use of technology resources, including AI, by city employees are subject to disciplinary action up to and including termination. This would include acts or omissions resulting in damage or disruption to the city's network, systems, services or data; or other negative impact to the city and/or its reputation.
- Suspected violations of this policy should be reported to the employee's supervisor, department director or the Human Resources director, depending on the severity of the violation.
- Violations of policy by third parties while operating an AI system on behalf of the City of Bellevue may result in breach of contract or termination of the contract and/or pursuit of other legal remedies.
- All departments share responsibilities for enforcing this policy.
Exceptions
Any exceptions to this policy must be documented, reviewed, and approved in writing by the city’s chief information officer (or equivalent authority) and must include an expiration date and compensating controls.
Document Changes and Feedback
This policy will be updated and reissued at least annually to reflect, among other things, changes to applicable law, update or changes to the City of Bellevue’s requirements, technology and the results or findings of any audit.