AI Tools Usage Guidelines for University Employees

The Annenberg School for Communication has established guidelines for the responsible and effective use of artificial intelligence tools. These guidelines address available AI services, appropriate use of University data, and best practices for faculty and staff.

Employees should review the complete Annenberg guidelines before using an AI tool for University work:

AI Tools Usage Guidelines for the Annenberg School for Communication

Using AI Tools with University Data

The information that may be entered into an AI tool depends on:

• Penn’s classification of the information as Low, Moderate, or High Risk; and

• the contractual and technical protections provided by the particular AI service.

Logging into a service with a Penn email address does not necessarily mean that the service is approved for University data. For contracted services, users must access the tool through the appropriate University account to receive Penn’s contractual protections. (ISC)

For definitions and examples of Penn’s data classifications, see:

Penn Data Risk Classification

Penn classifies University information as Low, Moderate, or High Risk based on its sensitivity, applicable laws and regulations, and the possible harm caused by inappropriate access or disclosure. (ISC)

Penn-Contracted AI Tools

Penn and ASC provide access to several AI tools with enhanced contractual and data protections. The types of University data permitted vary by tool.

Because available tools, pricing, and approved data classifications may change, employees should consult ISC’s current resource page:

Penn Generative AI Tools and Resources

Employees should not assume that every paid, enterprise, or institutionally branded AI account has been approved for University data. Contact ASC IT when the status of a tool or proposed use is unclear.

Public and Non-Contracted AI Tools

Do not enter Moderate-Risk or High-Risk Penn data, confidential or proprietary information, intellectual property, or unpublished research into a public or non-contracted AI service unless the use has been appropriately reviewed and authorized.

As a general rule, do not enter information into a non-contracted commercial AI service that you would not want disclosed publicly.

Penn’s central guidance discusses privacy, intellectual property, research data, accuracy, transparency, bias, and other considerations associated with generative AI:

Penn Guidance on the Use of Generative Artificial Intelligence

Penn notes that generative AI technology, vendor relationships, terms of service, regulations, and institutional practices continue to evolve. (ISC)

Best Practices

When using AI tools for University work:

• Verify AI-generated information before relying on or distributing it.

• Consider the privacy and sensitivity of all information entered.

• Follow applicable research, sponsor, contractual, and data-use requirements.

• Maintain academic and professional integrity.

• Document the use of AI when appropriate.

• Seek guidance before using AI with research data or other information that may not be Low Risk.

Questions and Access Requests

For help selecting an appropriate AI tool, requesting access, or evaluating a proposed use, contact the ASC IT Help Desk at ithelpdesk@asc.upenn.edu.