18.A.
9 Austin JAustin JAustin JAustin J, et al, et al, et al, et al. . . . Trauma Surg Acute Care OpenTrauma Surg Acute Care OpenTrauma Surg Acute Care OpenTrauma Surg Acute Care Open 2026; 2026; 2026; 2026;11111111:e002099. doi:10.1136/tsaco-2025-002099:e002099. doi:10.1136/tsaco-2025-002099:e002099. doi:10.1136/tsaco-2025-002099:e002099. doi:10.1136/tsaco-2025-002099 Open access
a 10- year age and sex- specific analysis of national injury data. Phys Sportsmed 2025;53:345–52.
5 Burford KG, Itzkowitz NG, Rundle AG, DiMaggio C, Mooney SJ. The Burden of Injuries Associated With E- Bikes, Powered Scooters, Hoverboards, and Bicycles in the United States: 2019‒2022. Am J Public Health 2024;114:1365–74. 6 Goodman LF , Birnbaum Flyer Z, Schomberg J, Maginas M, Wallace E, Vukcevich O, Awan S, Gibbs D, Nahmias J, Guner YS. Electric bicycles (e- bikes) are an increasingly common pediatric public health problem. Surg Open Sci 2023;14:46–51. 7 Maa J, Doucet JJ, Ignacio RC, Alfrey EJ. Electric Bikes Are Emerging as Public Health Hazard. Bull Am Coll Surg 2024;109:34–40. Available: https://www.facs.org/for- medical-professionals/news-publications/news-and-articles/bulletin/2024/julyaugust- 2024-volume-109-issue-7/electric-bikes-are-emerging-as-public-health-hazard/. 8 Younes H. Comparing injuries from e- scooters, e- bikes, and bicycles in the United States. J Cycling Micromobil Res 2025;4:100061.
9 United States Department of Transportation National Highway Traffic Safety Administration. MMUCC guideline model minimum uniform crash criteria. 5th edn.
Washington, DC, 2017.
10 King TL, Thornton LE, Bentley RJ, Kavanagh AM. The Use of Kernel Density Estimation to Examine Associations between Neighborhood Destination Intensity and Walking and Physical Activity. PLoS One 2015;10:e0137402.
11 Doucet JJ, Godat LN, Kobayashi L, Berndtson AE, Liepert AE, Raschke E, Denny JW, Weaver J, Smith A, Costantini T. Enhancing trauma registries by integrating traffic records and geospatial analysis to improve bicyclist safety. J Trauma Acute Care Surg 2021;90:631–40.
12 Frye W, Chehab L, Feler J, Wong L, Tan A, Alpers B, Patel D, von Hippel C, Sammann A.
Popular but precarious: low helmet use among shared micromobility program riders in San Francisco. Front Public Health 2024;12:1477473.
13 Cai J, Wang Y , Wang X, Zhang R, Li G, Guo F . Evaluation of risk perception among young electric bike riders. Ergonomics 2025;2025:1–14.
14 Kroes TR, Poos HPAM, Buijze ALF , Rozema R, Lameijer H, Emous M. Injury severity and traumatic brain injuries in youth and adolescents using conventional and electronic bicycles; a retrospective cohort study. Eur J Trauma Emerg Surg 2025;51:268.
15 Safe RIDE Act, H.R. 5265, 118th Cong., 2nd Sess; 2025. 16 Kobayashi LM, Williams E, Brown CV , Emigh BJ, Bansal V , Badiee J, Checchi KD, Castillo EM, Doucet J. The e- merging e- pidemic of e- scooters. Trauma Surg Acute Care Open 2019;4:e000337.
17 Black KM, Santorelli JE, Costantini TW, Kobayashi LM, Doucet JJ, Haines LN. From Peril to Protection: an evaluation of regulations impacting eScooter injuries. Trauma Surg Acute Care Open 2025;10:e001440.
18 PeopleForBikes Staff. The E- bike problem is an E- moto problem. 2025. Available:
https://www.peopleforbikes.org/news/the-e-bike-problem-is-an-e-moto-problem [Accessed 19 Sep 2025].
19 Agrawal AW, Fang K. Exploring e- bike safety performance data and policy options for California. Report No.: MTI Project 2423. San José (CA) Mineta Transportation Institute, San José State University; 2025. Available: https://transweb.sjsu.edu/sites/ default/files/2423-Agrawal-Electric-Bicycle-Safety-Data-Policy.pdf
18.A.
City of Hemet
City Council Meeting
STAFF REPORT
Meeting: City Council Meeting - Sep 22 2026
Department: City Attorney
Submitted by: Noah Rau, Interim City Manager
Steven Graham, City Attorney;
☐ Receive and File ☐ Consent Calendar ☒ Business Item ☐ Public Hearing
TITLE:
City Council Policy Direction Regarding Development of a Citywide Generative Artificial Intelligence Use Policy
RECOMMENDED ACTION:
It is respectfully recommended that the City Council:
1. Receive and file this report concerning the potential benefits and areas of concern associated with the use of generative artificial intelligence by City employees;
2. Provide policy direction to staff on the items presented under Requested Policy Direction below; and
3. Direct staff to prepare a draft Generative Artificial Intelligence Use Policy consistent with that direction and to return the draft to the City Council for consideration and adoption following information technology review, legal review, and any required notice and meet-and-confer process with the City’s recognized employee organizations.
This item is presented for direction only. No policy, resolution, or ordinance is presented for adoption at this meeting. This item approves no artificial intelligence product, subscription, expenditure, integration, or workflow, and establishes no new employment requirement. Staff has intentionally brought the underlying policy questions to the City Council before preparing a policy document, so that the Council’s direction shapes the draft rather than being sought after the fact.
EXECUTIVE SUMMARY:
Generative artificial intelligence refers to computational systems that produce text, images, audio, video, software code, or similar content in response to instructions or other inputs. In a municipal setting these systems reach employees in two distinct ways. The first is the familiar one: a standalone service that an employee opens deliberately. The second is less visible and, for purposes of this report, more consequential: generative features enabled inside software the City already licenses, including office productivity suites, records and agenda systems, permitting and code enforcement platforms, help desk systems, and transcription tools. Features of that kind can arrive through a routine vendor update without any City decision having been made about them.
18.B.
The artificial generative whether is therefore, not City the before question practical Council, intelligence will be present in the City’s software environment. It is whether the City establishes the conditions under which employees may use it, and who decides those conditions.
Existing City framework. The City’s Personnel Rules and Regulations already address electronic communications and privacy rights and workplace inspections and searches, and those provisions apply to City business conducted through artificial intelligence tools as they do to other electronic activity. The Police Department’s 2026 Policy Manual separately contains Policy 348, Generative Artificial Intelligence Use, and Policy 323.7, addressing artificial-intelligence-assisted reports, each administered through the Chief of Police. Outside the Police Department, the City has no policy specifically governing employee use of generative artificial intelligence.
General technology and acceptable-use rules were written for tools that store, transmit, and display information. They do not readily answer the questions a generative tool presents: whether a particular tool may be used for a particular purpose, what information may be placed into it, what verification the employee owes before relying on its output, and who remains accountable for the resulting work.
BACKGROUND/ANALYSIS:
Potential benefits. Used within defined limits, generative artificial intelligence can assist City personnel lengthy plans, reports, summarizing and and drafting preliminary with editing; correspondence; supporting translation and language access; organizing and analyzing data the City already holds; preparing routine constituent communications; and technical assistance with spreadsheets, forms, and code. For an organization carrying a substantial workload with a limited number of positions, the more realistic benefit is capacity relief on routine work rather than the reduction of positions or of professional judgment.
Staff notes two qualifications. First, the benefits described above are real but are not presently quantified for this City, and the experience of other agencies is not directly transferable. Second, anticipated efficiency gains should not be recorded as budgetary savings without evidence. If the Council authorizes a policy, staff would expect to measure actual results before representing savings of any kind.
Areas of concern. The same capabilities create risks that the City would need to manage. The principal areas are summarized below.
Accuracy and fabrication. These systems generate fluent, confident text that can be incorrect.
They can invent case citations, statutory provisions, quotations, figures, and source documents that do not exist. The error is not obvious on the face of the output, and it is not corrected by a signature. A policy would need to require verification of material facts, citations, and calculations against actual sources before reliance or release, with the depth of review scaled to the consequences of the work.
Confidential information. Information placed into a prompt, or reachable through a connected repository, may leave the City’s control unless addressed by the enterprise subscription and a vendor setting that ensures City inputs will not be accessed by third parties or used to train a model.
Bias and decisions affecting rights. Where these tools are used to support decisions about employment, benefits, permits, licensing, enforcement, or access to services, they can carry forward discriminatory assumptions from their training data and produce outcomes the City cannot explain. California employment regulations now address automated decision systems used in employment decisions and impose record-retention obligations for the underlying data.
The City would need to identify where a tool may and may not be used to support a determination affecting a person’s rights or significant interests.
18.B.
Accountability transparency. public and actually obscure can output who Generative exercised judgment on the City’s behalf. Related questions include whether and when the public should be told that artificial intelligence assisted in preparing a City communication, how realistic synthetic images, audio, or video depicting identifiable persons or events should be labeled and limited, and whether the City should offer a public-facing interactive service, which presents distinct questions of accuracy, accessibility, language access, and the availability of a human alternative.
Employment and labor effects. Some provisions of an artificial intelligence policy implement existing law or managerial prerogative. Others may materially affect represented employees’ working conditions, evaluation, discipline, monitoring, workload, or required training. Those effects may carry consultation and bargaining obligations d and they are a reason to obtain Council direction on the policy’s shape before a draft is circulated.
Procurement, cost, and unapproved use. Free tiers, personal subscriptions, browser activated review, purchasing without be features embedded and trials, extensions, can contract terms, or security evaluation. Unapproved use is the most likely path by which nonpublic City information reaches an unvetted vendor. Conversely, approved use is not costless: secure service tiers, configuration, accessibility, training, administration, and monitoring all consume funds or staff time.
Police Department requirements. Law enforcement use is subject to separate statutory and departmental requirements, including disclosure, officer verification, retention of the first artificial-intelligence-generated draft, and audit trails for covered reports, together with criminal justice information security obligations. Any citywide policy would supplement, and would not replace or relax, the Chief of Police’s approval authority and the Department’s existing requirements.
DISCUSSION:
Staff requests direction on the five questions below. Each is presented because it reflects a policy judgment properly made by the City Council rather than an operational or technical detail. Matters such as which specific products are evaluated, how accounts are configured, what the request forms look like, and how training is delivered are administrative and would be handled through the City Manager’s Office and Information Technology under whatever framework the Council directs. The Council may adopt the options as presented, modify them, or direct a different approach.
Whether to authorize employee use of generative artificial intelligence, and how broadly to begin. The threshold question is whether the City will permit employees to use these tools for City business and, if so, at what scale the City starts. Because generative features are already appearing in licensed City software, a decision not to address the subject is itself a decision about what employees may do.
Option A. Authorize development of a citywide policy permitting defined, approved uses, with implementation phased so that low-risk work involving public information begins first and more sensitive uses follow as safeguards are verified. (Suggested by staff.)
Option B. Authorize a limited pilot in identified departments under the same safeguards, with a report to the Council before any broader implementation.
18.B.
Option C. Direct that employee use be prohibited pending further study, recognizing that existing embedded Police the and restricted, and that be to need would features identified Department’s separate requirements would continue to govern its uses.
The approval architecture, and what remains with the City Council. Any workable policy must distribute decisions among the Council, the City Manager, Information Technology, department heads, and reviewing offices. The policy question is how much the Council delegates and what it keeps.
Option A. Adopt a risk-tiered framework in which routine low-risk assistance is authorized without case-by-case approval, sensitive uses require documented approval by the department head and Information Technology with specialist review, and the most sensitive uses additionally require the City Manager and the City Attorney. The Council would reserve to itself material changes to the policy’s coverage, its prohibited uses, permissions for protected information, approval authority, retention duties, and provisions affecting employee rights, and would receive an annual report. (Suggested by staff.)
Option B. Require City Council approval of each artificial intelligence tool and each category of use before it may be employed.
Option C. Delegate administration of the subject to the City Manager in full, with an annual informational report to the Council.
Uses the City will categorically prohibit. Certain conduct is unlawful regardless of policy. Beyond that, the Council may establish City prohibitions that no administrator may waive. Staff recommends that the following be prohibited as a matter of City policy, subject to the Council’s direction to add to, narrow, or decline any of them: making or effectively determining a final decision affecting a person’s rights or significant interests; ranking or scoring identifiable applicants or employees for hiring, promotion, discipline, or termination; inferring credibility, emotion, character, protected status, or criminal propensity from appearance, voice, or biometric indicators; covert or continuous monitoring of employee behavior; operating a tool with uncontrolled authority to communicate, alter records, spend funds, or transact on the City’s behalf; and entering nonpublic City information into an unapproved tool or a personal account. A corresponding question is whether a change to any City prohibition should require a return to the Council, as staff recommends, rather than administrative amendment.
Transparency to the public, synthetic media, and public-facing services. Three related questions are presented. First, when should the City disclose that artificial intelligence assisted in preparing a City work it analysis, substantive supplied where material instance, every in product: only recommendations, or disputed facts, or only where disclosure is required by law? Staff suggests the middle course, coupled with disclosure whenever its absence would leave a materially misleading impression. Second, should realistic synthetic images, audio, or video depicting identifiable persons or clear a and heightened basis, lawful approval, with only permitted be events actual a contemporaneous label? Third, should the City permit a public-facing interactive service at all, and if so, only on conditions that it identify itself as automated, be limited to identified City information, provide a readily available path to a City employee, and meet accessibility and language access requirements before launch rather than after?
and reporting. expectations, Resources, training, workforce Council’s the Finally, is direction requested on the posture toward implementation costs and workforce effects. Should the policy be developed and implemented within existing budget and staffing, with any additional appropriation
18.B.
requested separately if it proves necessary, as staff suggests, or should staff return with a costed implementation plan and a proposed appropriation? Should the policy make it clear that these tools are intended to assist City personnel rather than to displace them, and that projected efficiency gains will not be recorded as savings without evidence? Should training and a written acknowledgment be required before an employee receives access? And should the City Manager report implementation status to the Council within 90 days of adoption and thereafter at least annually, as staff suggests?
STRATEGIC PLAN INTEGRATION:
GOAL FIVE: ORGANIZATIONAL EFFECTIVENESS
To identify the work needed inside the City organization to effectively provide services to the community and ensure financial stability.
Objective 1: Operate in a financially sustainable manner.
Objective 2: Evaluate existing practices to ensure continuity with industry best practices.
Objective 3: Provide Council with policies that are consistent while lessening liabilities.
Objective 4: Increase employee retention by training and creating a positive workplace.
Objective 5: Streamline fire prevention coordination efforts.
18.B.