Loading...
HomeMy WebLinkAboutAdministrative Order 2024-08 Adoption of General AI PolicyAppendix D- Generative Artificial Intelligence (AI) Use Policy Section 1: Purpose and Scope The purpose of this policy is to provide guidance to City of Bozeman (City) employees regarding the use of Generative AI in the performance of their job duties. Generative AI is defined as a deep learning model that can generate high quality content such as stories, images, voice replication, and music. Examples of Generative AI include but are not limited to: ChatGPT, DALL-E, and Google Gemini. For the purposes of this policy, “employee” shall include all City employees and volunteers. Generative AI may, in some cases, be a useful tool for employees and may help increase productivity. Due to the potential increase in productivity, the City will authorize the use of Generative AI in accordance with this policy. Generative AI has developed rapidly and is embedded in many applications commonly used in an office setting. Due to this rapid development, employees authorized to use Generative AI are expected to routinely check for any updates to the City’s Generative AI Use Policy and FAQ sheet issued by the City IT department, to ensure they remain up to date on current standards. Employees are responsible for their own work product, including any work product created with the assistance of Generative AI. If authorized to use Generative AI in the course and scope of their employment, employees must be mindful that plagiarism is not acceptable. This policy is not intended to apply to integrated AI tools such as using a Google search function. Section 2: Director Authority Each department director must address this policy with their staff. The IT Director has general discretion over which Generative AI platforms are appropriate for employees to use, and each department director has discretion over whether the use of Generative AI is allowed for their employees. Department directors may only authorize their employees to use a Generative AI system that has been approved by the IT Department. A director may deny the use of Generative AI within their department at any time. Section 3: Employee Expectations Upon approval by their director, employees may be permitted to use Generative AI in the course and scope of their City employment. Employees authorized to use Generative AI in the course and scope of their employment must do so on a City account, accessible to the IT Department. Prompts inserted into Generative AI and the results given may be considered public information and subject to record retention requirements and disclosure. Therefore, it is necessary for authorized employees to use a City account to ensure compliance with record retention policies. Employees who are authorized to use Generative AI must abide by the following: • Accuracy: Employees are ultimately responsible for the content of their work. Should an employee use Generative AI in the course and scope of their employment, the employee must fact check any results produced for accuracy. Importantly, Generative AI may only contain information up to a certain date and therefore can produce an output that may be outdated or simply incorrect. • Confidentiality: Confidential information and personal identifiable information (PII) must never be shared with Generative AI. Since Generative AI is a relatively new tool, questions still exist about who owns data once it is submitted to Generative AI. It is possible that once data is submitted, the Generative AI system is the owner of the data. Further, it is difficult to know what happens to data once it is submitted to Generative AI. It is possible that confidential information could be released or shared with a third party. Additionally, entering PII into Generative AI may breach privacy laws such as HIPAA. • Transparency Employees must be transparent if using Generative AI and must be able to demonstrate to their supervisor the extent to which Generative AI is used in the course and scope of their employment. Employees must cite to Generative AI on any work product that requires citation, as directed by their supervisor. • Bias Generative AI creates content based on patterns and relationships the AI system learns from existing human knowledge. This means content created by Generative AI may have the same implicit bias that humans have. Employees using Generative AI cannot trust that the output provided is accurate and free of bias. Employees are responsible for any work product produced by Generative AI and as such must reconcile any bias that may exist. • Copyright If an employee is authorized to use Generative AI in the course and scope of their employment, the employee must ensure material is not subject to copyright, as Generative AI has the potential to infringe on a copyright owners’ rights. Section 4: Vendors and Contractors City vendors and contractors must abide by the terms of their contract or agreement with the City regarding their use of Generative AI on City projects. The City’s most recent template professional services agreement includes a requirement for contractors to disclose the use of Generative AI. Section 5: Training Any director who plans on authorizing the use of Generative AI in their department and any employee who plans on using Generative AI on City work product must complete a training on appropriate use of Generative AI approved by the IT Department. Section 6: Violations Any violation of this Policy is subject to Section 10 of the Information Technology Use Policy.