Generative AI guidelines for SFSU employees

Introduction

San Francisco State University seeks to explore and appropriately use generative artificial intelligence (GenAI) technology within the institution. The technology may be transformative in business, academic, and practice areas but is quickly evolving and carries risks for which users must be aware as they evaluate and integrate GenAI into their work at SF State.

The availability of GenAI within campus tools and services is dependent upon the agreement of the service providers, the Chancellor’s office, and the campus. Please visit the SF State AI website to learn more about tools available to the campus. As with all technology, the best practices cited in this document and in campus training can guide responsible and effective use by campus employees.

About these guidelines

These guidelines focus on the general use of GenAI by SF State employees (including staff, administrators, and faculty), encouraging responsible and effective use at SF State. Detailed training on the guidance in this document is offered through the campus AI Literacy Education Program.

This document does not address: (1) specific instructional guidance to faculty on teaching with AI or (2) student-specific guidance related to appropriate use of AI – which may be developed in the future with the support of campus governance. This is not a policy document, but highlights existing California State University (CSU) and SF State policies that are relevant to the use of GenAI (refer to table below).

Policy

Area of Observance
CSU Information Security Responsible Use Policy:  https://calstate.policystat.com/policy/14685892/latest/ Users should help maintain secure use of software, ensure their own personal data privacy, and not violate other state and CSU regulations.
SF State Sensitive Data Policy: For definitions of Sensitive Data (Confidential, Level 1, Level 2) please refer to the University Confidential Data practice directive: https://qaservices.sfsu.edu/confidential-data Do not upload or input sensitive data into GenAI tools or platforms. For projects or tasks that may require the upload of sensitive data, consult with an IT provider.

SFSU Academic Integrity Policy (#S22-298): https://sfsu.policystat.com/policy/11521145/latest

&

Spring 2024 Resolution (#RS24-466):
https://sfsu.policystat.com/policy/16405204/latest

 Instructors can refer to Academic Senate academic integrity policy and the Spring 2024 Resolution on the Use of GAI in Teaching and Learning for guidance on teaching or student integrity concerns.
San Francisco State University Brand Guidelines
https://marcomm.sfsu.edu/brand		 Guidelines for representing the University brand in the creation and sharing of information assets or publications should be followed. AI-generated assets should not be used to represent SF State’s brand.

As GenAI evolves, these guidelines will be periodically reviewed and updated. 

Feedback?

The SFSU AI team wants to hear from you. 

Provide any feedback, thoughts, or suggestions via email (ai@sfsu.edu).

Provide feedback on the guidelines

View and download file (.pdf)

Guidance for Using Generative AI at San Francisco State University

The recommendations listed below guide employee use of GenAI technology as it may be used in work and other activities at SF State. Campus GenAI training can reinforce this guidance in greater detail. Please review available AI trainings offered through the AI Literacy Education Program.

  • Supplement to Human Work: GenAI is intended to support and enhance human work, not replace it. GenAI should serve as a complement to work, and employees should maintain an active role in the creation of work products and should carefully evaluate it. Human work plays a critical role in maintaining accuracy, fairness, and integrity. The use of GenAI in work should be reviewed with an appropriate administrator. GenAI output should be referenced in work products through citation or notes (refer to section below). Output generated by GenAI tools or platforms should be cross-checked and corroborated.
  • Professional Development in GenAI Proficiency: Users should strive to enhance their knowledge and skills in using GenAI technologies. GenAI proficiency ensures informed use and maximizes the benefits of AI. Users should take advantage of campus-based or CSU GenAI trainings that educate users on basic awareness, hands-on use, and critical AI literacy (find training opportunities at https://ai.sfsu.edu/ai-events).
  • Data Privacy and Security:  Sensitive University data can be put at risk when using GenAI tools (refer to Sensitive Data policy https://qaservices.sfsu.edu/confidential-data). Sensitive or FERPA-related university data or materials should not be uploaded or input into any GenAI system (including the campus licensed such as Microsoft Copilot). GenAI tools ingest data into their training model putting sensitive data at risk. Before using GenAI tools, consult your unit’s IT provider and appropriate campus administrator on proposals or projects that may benefit from the inclusion or use of sensitive data generated by the University or on projects impacting confidential campus processes or critical decision-making, such as recruitment and hiring.
  • Accuracy and Quality:  GenAI tools can produce inaccurate, outdated, biased or fabricated output called “hallucinations.” Employees are responsible for ensuring content created with the support of GenAI is accurate and free from errors or bias. Units seeking to use GenAI in their processes should consider review and approval steps to verify content. Use of GenAI in any confidential or critical campus process which may generate liability or similar issues requires prior consultation with your IT provider, as well as other campus administrators (Human Resources, Risk Management, ORSP, Faculty Affairs, etc.).
  • Bias In GenAI Output: GenAI data models and outputs often reflect societal biases related to gender, ethnicity, culture, and socioeconomic status. Training data is predominantly scraped from the internet, where these biases exist in human-authored content. GenAI tools are also created primarily in the US, with a Western frame of reference and little transparency in the development process. As a result, these tools may favor specific viewpoints. Users should critically evaluate whether diverse identities and experiences are represented in GenAI content, check carefully for stereotypes, and ensure reference sources offer balanced perspectives.
  • Citation and Attribution of GenAI Content: Employees should acknowledge the use of GenAI in University work or product that includes significant AI contributions. Document the source of GenAI content (for example: OpenAI. (2023). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/chat). Methods used to develop output should be explainable. Currently, formal citation for GenAI can be found via the Library guide on citation: https://libguides.sfsu.edu/AI/citation.
  • Copyright and Intellectual Property: Broad national concerns about whether GenAI tools are violating copyright law based on the training data used is still being determined. For SF State employees using GenAI in their work, copyrighted work such as writing, images, and other content should not be uploaded or input into GenAI tools without the owner's expressed permission. For more information, refer to the Library’s “Researching with Generative Artificial Intelligence” guide: https://libguides.sfsu.edu/AI/copyright.

 

Tools and Recommendations

The varieties of Generative AI tools are expected to change rapidly. Currently, chat-based GPT tools are among the most popular (including ChatGPT, Copilot, and Gemini). Campus licensed software platforms are releasing GenAI-based features (including Zoom, Adobe Creative Suite, and Box). We also expect more specialized applications and practices to evolve as GenAI technology advances. SF State will need to develop thoughtful approaches to new and changing capabilities.

Any software tool into which data or documents can be input, regardless of GenAI technology, requires adherence to data privacy and security measures by users. Tools licensed and under contract by the University are typically covered by CSU terms and conditions that protect how data can be used (but never input or upload sensitive data). Microsoft Copilot is one current licensed version of a chat-based GenAI tool available to all faculty, staff, and students. 

Please consult the SF State AI Website’s Tools page for the latest information on available GenAI software and how to request GenAI software.

 

Where to Get Help