UDAR Generative AI Recommendations and Guidelines

Overview

Generative AI (GenAI) can boost efficiency and creativity across many tasks. Used responsibly, it supports both operational and strategic work. However, benefits must be balanced with awareness of risks. Before using GenAI, understand potential challenges and follow UC risk mitigation policies.

The GenAI landscape is evolving quickly. As new capabilities and risks emerge, guidance will be updated. Staying informed and adaptable is key to safe, ethical, and effective use.

Getting Started

Complete the online course UC AI Primer to build shared foundational knowledge, regardless of experience level.
Before using any GenAI tool, review the maximum allowable data protection level.
Learn more about data privacy levels here.
Do not enter or generate confidential, proprietary, or sensitive information in GenAI tools, as inputs may be stored or reused by vendors.
Is there a GenAI tools or capabilities from existing tools you'd like to use? Submit a request through the Data and Technology Intake process.
All AI use must align with UC Berkeley’s Principles of Community and the UC Principles of Responsible AI (full report).

Do not use GenAI with UC branded visuals. Use official media only for UC Berkeley wordmark, Cal script logo, seal, campus images, students, etc.

Advancement Using GenAI

Improve productivity by utilizing GenAI responsibly and efficiently. Use cases and appropriate uses have been identified.

Reference Advancement's GenAI Guiding Principals

Sensitive Data

Our UC Berkeley P3 approved AI tools are designed with internal safeguards which allow for the usage of CADS data; however, they are not approved for the storage, processing, or analysis of highly sensitive donor information. To protect donor privacy and ensure compliance with University data governance standards, employees must exercise caution when using AI-assisted systems for work involving donor data.

Please use discretion and avoid entering any information that could compromise donor privacy or confidentiality.

Examples of information that should not be entered include:

Donor-disclosed health information
Political affiliations
Race or ethnicity data tied to individuals
Confidential records
Financial data (credit card, bank information, non-public company information)
Relationships (separations or divorces not yet announced)

When in doubt, use de-identified or generalized data. If you’re unsure whether something is appropriate to include, please check with the Advancement Data team before proceeding.

University Development and Alumni Relations