Loading...
HomeMy WebLinkAbout029-26 - Resolution - AI PolicyDocusign Envelope ID: 06448A97-BBFB-88EE-8389-D067A3C23C3C RESOLUTION NO. 029-26 A RESOLUTION OF THE CITY OF PORT ORCHARD, WASHINGTON, ADOPTING A MUNICIPAL ARTIFICIAL INTELLIGENCE POLICY TO GOVERN THE RESPONSIBLE ACQUISITION AND USE OF Al TECHNOLOGIES BY CITY DEPARTMENTS AND EMPLOYEES. WHEREAS, the City of Port Orchard recognizes that Artificial Intelligence (Al) technologies are rapidly evolving and offer significant opportunities to improve public services and advance beneficial outcomes for the community; and WHEREAS, the City is committed to the principles of transparency, accountability, and data privacy in the deployment of new technologies; and WHEREAS, the City acknowledges the potential for Al systems to perpetuate bias or inequity and seeks to implement formal "Human -in -the -Loop" (HITL) requirements to ensure accuracy and fairness in Al -generated outputs; and WHEREAS, a City Al Committee has developed a comprehensive policy to provide a framework for the effective and responsible use of Al systems, including specific guidelines for data classification and public records management; WHEREAS, the IT Department and the Al Committee have established an approval process for the acquisition of Al tools to ensure they meet the City's security and resiliency standards; now, therefore, THE CITY COUNCIL OF THE CITY OF PORT ORCHARD, WASHINGTON, HEREBY RESOLVES AS FOLLOWS: THAT: The "City of Port Orchard Artificial Intelligence Policy," attached hereto as Exhibit A, is hereby adopted to be effective immediately. All City departments, employees, elected officials, and contractors engaged with City data are bound by the requirements and principles set forth therein. PASSED by the City Council of the City of Port Orchard, SIGNED by the Mayor and attested by the City Clerk in authentication of such passage this 28th day of April 2026. Signed by: I�o� �v�a.atn,suu. Robert Putaansuu, Mayor ATTEST: Signed by: Brandy Wallace, MMC, City Clerk 0 i44 t.sEA:J Docusign Envelope ID: 06448A97-BBFB-88EE-8389-D067A3C23C3C CITY OF PORT ORCHARD 910 - Artificial Intelligence Policy Effective Date: ASAP PURPOSE The purpose of this policy is to set forth the requirements to observe when acquiring and using software that meets the definition of "artificial intelligence." SCOPE All individuals and entities (herein defined as "Users"), including City departments, employees, elected officials, vendors, contractors, and volunteers, who operate under the authority of the City of Port Orchard and engage with City data are bound by this policy. DEFINITIONS Terms used in the current Artificial Intelligence space are fluid and dynamic. Attached in Appendix A is a list of current definitions that have been approved by the Al committee to ensure a shared understanding of its scope and application. ARTIFICIAL INTELLIGENCE (Al) PRINCIPLES These Principles describe general codes of conduct that represent the values and responsibilities of the City to its residents. This policy serves to inform Users in their use of Al technology. Users shall adhere to the principles and requirements outlined in this policy. • Innovation: The City values public service innovation to meet our residents' needs. We commit to responsibly explore and continuously evaluate Al technologies, which will improve our services and advance beneficial outcomes for our community. • Transparency and Accountability: The City values transparency and accountability and understands the importance of these values in our use of Al systems. The City will ensure that the development, use, and deployment of Al systems are evaluated for and compliant with all laws and regulations applicable to the City prior to use and will make documentation related to the use of Al systems available publicly. • Validity and Reliability: The City will work to ensure that Al systems perform reliably and consistently under the conditions of expected use, and that ongoing evaluation of system accuracy throughout the development and/or deployment lifecycle is managed, governed, and auditable. • Bias, Harm Reduction, and Equity: The City acknowledges that Al systems have the potential to perpetuate inequity and bias resulting in unintended and potentially harmful outcomes. The City will evaluate Al systems with a strong focus on equity, addressing potential impacts arising from data, human, or algorithmic bias. • Data Privacy: The City values data privacy and understands the importance of protecting personal data. The City strives to ensure that policies and standard operating procedures Docusign Envelope ID: 06448A97-BBFB-88EE-8389-D067A3C23C3C reduce privacy risks, and are applied to Al systems throughout development, testing, deployment, and use. • Explainability and Interpretability: The City understands the importance of leveraging Al systems, models, and outputs that are easily interpreted and explained. The City will attempt to ensure all Al systems utilized, and their outputs, are communicated in clear language, representative of the context in which they are deployed. • Security and Resiliency: Securing our data, systems, and infrastructure is important to the City. The City will ensure Al systems are evaluated for resilience and can maintain confidentiality, integrity, and availability of data for critical City systems. The City will actively work to minimize security risks in alignment with governing policy and identified best practices. POLICY 1. Acquisition and Usage of Al Technology 1.1. The City has a list of approved Al tools that can be used in accordance with individual department policy (see Appendix B). This list applies to all employees of the City and elected officials. 1.2. Consistent with the City's standards for Acquisition of Technology Resources, Department directors may request acquisition of Al tools (not listed in Appendix B) through the City's current IT request process. Departments may not acquire or use Al systems without following the approval process identified in this policy. 1.3. The IT Department shall review requests according to its current risk and impact methodology, which shall include specific review criteria for Al technology (see Port Orchard Acquisition of Al Review Guidelines). The IT Department will then bring the request to the Al Committee for approval or denial. 1.4. The City's standard for technology acquisition applies to all technology, including Open Source, Free to Use software, or SaaS (Software as a Service) tools. 1.5. If a technology that has already been approved for use in the City adds or incorporates Al capabilities, the IT Department shall immediately issue a technical restriction to suspend the use of those new Al capabilities pending formal review. The Al Committee will immediately be notified and will then evaluate the change to ensure it continues to follow this policy, according to its current risk and impact methodology (see Section 1.3). If the Al Committee approves the new Al capabilities, the IT Department may lift the technical restriction. 1.6. The City's IT Department shall revoke authorization for a technology that adds Al capabilities and/or restrict the use of those Al capabilities; if those Al capabilities present risks that cannot be effectively mitigated to comply with this policy or other City policies until the risks can be addressed. • The IT Department's responsibilities include the technical authority to implement immediate and mandatory suspension or restriction of new Al capabilities added to existing software, as required by Section 1.5. The IT Department is responsible for notifying the impacted department(s) of the suspension and/or restrictions. 2. Use of Al Outputs Docusign Envelope ID: 06448A97-BBFB-88EE-8389-D067A3C23C3C 2.1. Outputs of Al systems shall be reviewed by human(s) prior to each use in an official City capacity ("Human in the Loop" or HITL). 3. Attribution, Accountability, and Transparency of Authorship 3.1. Any outputs generated by an Al system, and used substantively in a final product, requires attribution to the relevant Al system used. • Departments shall interpret substantive use thresholds to be consistent with the principles outlined in this document, as well as relevant intellectual property laws. • All attributions should include the name of the Al system used plus an HITL assertion (which should include the department or group that reviewed/edited the content). **Example** "Some material in this brochure was generated using ChatGPT 4.0 and was reviewed for accuracy by a member of the Department of Human Resources before publication." 4. Reducing Bias and Harm 4.1. Al systems may produce outputs based on stereotypes or use data that is biased. Users will review and evaluate Al generated content and the designated HITL reviewer shall ensure that the output is accurate and free of discrimination and bias (HITL). 5. Data Privacy 5.1. Use of Al tools shall be consistent with the principles and standards described in the City's Data Privacy Policy, and Information Security Policy. 5.2. Unless suitable enterprise controls and data protection mitigations are in place, users shall not submit any of the following to Al systems outside of the City's control: • Data classified by the City's data classification guidelines as "Confidential" or "Highly Confidential" as listed in the Port Orchard Data Classification Guidelines. • Data that constitutes a Protected Data or Public Record that is exempt from disclosure under the Revised Code of Washington (RCW) (RCW 42.56, et al.) or other applicable laws. • Any other non public information such as: preliminary drafts, intra-agency memorandums, or sensitive operational data that is not officially approved for public disclosure. • Data that is not considered to be acceptable to disclose to the public 5.3. No City data or records, including inputs or prompts, are to be used for training or parameter tuning for Al models outside the City's control. Al technologies that cannot prevent City data or records from contributing to their language models may not be used. 6. Public Records & City Records Management 6.1. All records generated, used, or stored by Al vendors or solutions may be considered public records and must be disclosed upon request. Docusign Envelope ID: 06448A97-BBFB-88EE-8389-D067A3C23C3C 6.2. All Al solutions and/or vendors approved for City use shall be required to support retrieval and export of all prompts and outputs (either via exposed functionality or through vendor contract assurances). 6.3. Users who use Al tools are required to maintain, or be able to retrieve upon request, records of inputs, prompts, and outputs in a manner consistent with the City's records management and public disclosure policies and practices. EXCEPTIONS Any exceptions to this policy must be approved in advance through submission to the City Al Committee. POLICY COMPLIANCE Noncompliance may result in the Mayor, department directors, or their designees, imposing disciplinary action, up to and including termination of employment or vendor contract. RELATED STANDARDS AND POLICIES • 901 - City of Port Orchard IT Data Security Policy • 902 - City of Port Orchard IT Data Privacy Policy • City of Port Orchard Acquisition of Al Review Guidelines • City of Port Orchard Data Classification Guidelines • National Institute of Standards and Technology Artificial Intelligence Risk Management Framework (NIST Al 100-1) (In Draft Folder) • Revised Code of Washington (RCW): o Ethics In Public Service (RCW 42.52) o Public Records Act (RCW 42.56) o Preservation and Destruction of Public Records (RCW 40.14) Docusign Envelope ID: 06448A97-BBFB-88EE-83B9-D067A3C23C3C RESPONSIBILITIES This policy will be maintained through the City's Al Committee. Their responsibilities include creating and maintaining the Al risk and impact criteria, developing mandatory standards for the uniform interpretation and measurement of "Substantive Use" (Section 3.1), criteria by which to evaluate suggestions for amendments to Appendix B, (see City of Port Orchard Al Review and Guidelines Policy), including the creation of mandatory departmental standards for designating and training HITL reviewers. DOCUMENT CONTROL This policy shall be effective on TBD and shall be reviewed regularly, or as appropriate. POLICY REVIEW AND APPROVAL HISTORY Version Content v 0.1 Initial Draft v 0.2 2nd Draft v 0.5 5th Draft v 0.6 6th Draft Contributors Reviewer(s): Noah Crocker -Finance Director, Rebecca Zick-Deputy Finance Director, Kori Pearson- Acct Asst III/IT Specialist, Alan Iwashita-POPD Deputy Chief, Sean Dunham -IT Manager, Jake Langston -IT Specialist, Caden Cucciardi- Asset Management Tech Reviewer(s): Noah Crocker -Finance Director, Rebecca Zick-Deputy Finance Director, Kori Pearson- Acct Asst III/IT Specialist, Alan Iwashita-POPD Deputy Chief, Sean Dunham -IT Manager, Jake Langston -IT Specialist, Caden Cucciardi- Asset Management Tech Reviewer(s): Noah Crocker -Finance Director, Rebecca Zick-Deputy Finance Director, Kori Pearson- Acct Asst III/IT Specialist, Debbie Lund — HR Director, Sean Dunham -IT Manager, Jake Langston -IT Specialist, Caden Cucciardi- Asset Management Tech, Jenine Floyd — Deputy City Clerk, Nick Bond — DCD Director Reviewer(s): Noah Crocker -Finance Director, Rebecca Zick-Deputy Finance Director, Kori Pearson -Acct Asst III/IT Specialist, Sean Dunham -IT Manager, Jake Langston -IT Specialist, Caden Cucciardi- Asset Management Tech Reviewer(s): Noah Crocker -Finance Director, Rebecca Zick-Deputy Finance Director, Kori Pearson- Acct Asst III/IT Specialist, Debbie Lund -HR Director, Sean Dunham -IT Manager, Jake Langston -IT Specialist, Caden Cucciardi- Asset Management Tech, Jenine Floyd —Deputy City Clerk, Nick Bond—DCD Director, Jim Fisk —Senior Planner, Alan Iwashita-Deputy Police Chief Reviewer(s): Noah Crocker -Finance Director, Kori Pearson- Acct Asst III/IT Specialist, Debbie Lund —HR Director, Sean Dunham -IT Manager, Jake Langston -IT Specialist, Caden Cucciardi- Asset Management Tech, Jenine Floyd —Deputy City Clerk, Nick Bond—DCD Director, Jim Fisk —Senior Planner, Alan Iwashita-Deputy Police Chief Approval Date July 17, 2025 August 21,2025 November 20, 2025 December 9, 2025 Docusign Envelope ID: 06448A97-BBFB-88EE-83B9-D067A3C23C3C Final v 1.0 Version Reviewer(s): Noah Crocker -Finance Director, Rebecca Zick-Deputy Finance Director, Kori Pearson- Acct Asst III/IT Specialist, Debbie Lund -HR Director, Sean Dunham -IT Manager, Jake Langston -IT Specialist, Caden Cucciardi- Asset Management Tech, Brandy Wallace -City Clerk Decemberl 6, 2025 Docusign Envelope ID: 06448A97-BBFB-88EE-83B9-D067A3C23C3C APPENDIX A: GENERAL Al DEFINITIONS Al (Artificial Intelligence): The capability of a machine or system to perform tasks that typically require human intelligence, such as generating text or audiovisual content, making or recommending decisions, analyzing data, or automating processes. Al Algorithm / Al Model: A set of programmed instructions that processes data to perform tasks, make decisions, or solve problems within an Al system. Al Assistant: An Al tool which is intended to aid a user in their day-to-day work by suggesting content, retrieving information, automating processes, and performing other similar tasks. Al Policy: A document that provides a framework for the effective and responsible use of Al systems within an organization. Al System: Any tool, software, process, and workflow, or other system which is based on Al technology, or which uses Al technology as a key component of that system. Al Tool: A piece of software which provides Al functionality and can be applied to a specific use case. Al Use Case: A specific task or purpose for which an Al tool is used or under consideration. Al User: An individual who is responsible for using, developing, purchasing, configuring, or maintaining Al systems. Al User Guide: A document which supplements an Al policy with more detailed guidance on how to implement the policy. Agentic Al: Autonomous Al systems that can act independently to achieve predefined goals, making decisions and taking actions without constant human oversight. Anonymization: A process by which data is altered so that it cannot be connected to specific individuals or organizations. Bias: Systematic tendencies that can exist within Al systems, often stemming from flawed data, algorithms, or design processes, that may lead to discriminatory or inaccurate outcomes affecting certain groups or individuals. Black Box Algorithm: An Al algorithm which produces decisions or other outputs with little or no mechanism for the user to analyze the logic which led to that result. Data Privacy: The protection of non-public information about a person or organization from disclosure without their consent. Data Output: Refers to the information produced by a computer or device as a result of processing input data. It can take various forms, including; Text, Images, Audio, Video, and more. Generative Al: Al systems which use algorithms to create text, audio, image, or video content based on some combination of user prompts and stored data and instructions. Human -in -the -Loop (HITL): A system design philosophy that requires human oversight and intervention in Al -driven processes. This ensures that final, critical decisions are made by a person, not the algorithm. Large Language Model (LLM): A type of generative Al that is trained on a massive amount of text data to understand, summarize, generate, and predict new human -like language. Machine Learning: A type of Al which uses algorithms to extract information from and recognize patterns in data, often used for forecasting, prediction, classification, or analysis. Open Data: Information derived from an organization's operations which is shared publicly to promote transparency or for use by external parties. Protected Data: Information generated or acquired in the course of an organization's operations which is not intended or approved for public disclosure. Docusign Envelope ID: 06448A97-BBFB-88EE-83B9-D067A3C23C3C Public Record: Records stemming from an organization's operations which must by law be shared upon request, whether or not it is actively published as open data. Docusign Envelope ID: 06448A97-BBFB-88EE-83B9-D067A3C23C3C APPENDIX B: Al TECHNOLOGY REGISTRY (Al Tools Currently Approved By Al Committee) Microsoft Copilot