Generative Artificial Intelligence Use Policy or Guidelines for Employees

Recognizing artificial intelligence’s (AI’s) potential to increase productivity, streamline work processes, and nearly eliminate routine repetitive tasks, your campus already may be using AI — specifically generative (GenAI) — in accounting, admissions, or even student success.

But not all GenAI tools are created equal, nor is GenAI appropriate for all work situations. To ensure appropriate workplace use, create a formal policy or, short of that, guidelines to help employees understand when and how to use AI on the job. Using these tools appropriately can help the institution avoid misuse and potential legal liability.

What Is GenAI?

Despite the hype, not everyone on campus knows about it. To help your staff identify when your policy or guidelines might apply, you may need to provide them fundamental education.

GenAI is a computer-based deep learning model that — after being trained on large sets of data — generates text, images, or other outcomes in response to specific requests or “prompts.” The final product is a result of the model’s ability to detect patterns in the large data sets it was trained on and use that information to produce something new and responsive to the prompt.

Create a Policy or Guidelines

Because of the novelty of these tools and their many uses, several institutions are struggling with whether to create a formal use policy for employees or merely release guidelines that incorporate promising practices while allowing users flexibility to determine when and how to use these tools on the job. Guidelines are generally voluntary, established to create consistency across campus, and not strictly enforced; a policy is mandatory and contains consequences for violations.

Make your decision to create a policy or release guidelines after seeking input from your Information Technology, Legal, Risk Management, and Human Resources departments. Some factors to consider while you’re deciding on creating a policy or releasing guidelines:

• Timeline to create a policy. You may want guidelines in place while you take the time and consideration to draft and approve a formal use policy.
• The desire for flexibility. Because this area is in flux, guidelines may be more easily updated and adjusted based on changes in technology and use cases.
• Different levels of usage among departments or jobs. Some staff only may use GenAI occasionally in their work while others will use it more heavily. You may need different levels of usage oversight based on how the tools are used.

Suggested Elements of a Policy or Guidelines

Once your institution has decided on issuing a policy or guidelines, inventory campus programs or products currently using GenAI. The objective is to understand what risks or concerns you need to address. As you draft your written guidance, consider including:

• Acceptable and unacceptable uses of GenAI. This may vary by department. You also may wish to name specific GenAI programs or products that your institution has approved, along with a list of those that have been prohibited, as well as how to see approval for new tools. Consider organizing your list of acceptable and unacceptable uses by perceived risk and severity of problematic usage.
• Prohibit using confidential information. Generally, because many of these products are “open” or public and continue to be trained on data input into the system, you should prohibit employees from putting confidential or personally identifiable information into these tools. Include examples of what is considered confidential data at your institution.
• Enumerate risks, including cybersecurity concerns, data privacy and protection (as noted above), and any intellectual property concerns that may arise. Include a person or department to contact if users have questions or concerns.
• Understand the terms and conditions of tools you use. Reiterate that tools used should be preapproved and employees shouldn’t agree to terms and conditions without properly understanding their impact on your institution, including legal ownership of input data and the tool’s output. Consider requiring legal review of terms and conditions of your approved GenAI tools.
• Require human oversight. The person using the tool should review all outputs from these tools for accuracy, including spot-checking of sources or facts. This review can help avoid spreading incorrect information.
• Determine if disclosures are required. Depending on the type of tool used and the work product, consider whether employees may be required to disclose their use of GenAI in its creation. This transparency will help others understand what has taken place and who or what exactly created the text or images being viewed.

Dissemination and Training

Once you decide on your approach and draft your policy or guidelines, inform employees. Widely distribute your document and post it in an accessible place online. Consider whether you need additional in-person or online training sessions, however brief, to drive home the policy message and implications.

Inform employees how to seek answers to questions they have or what they should do if there is a usage violation. Detail consequences for violating the use policy or guidelines.

This education helps people understand when these tools are best used and to try to avoid inappropriate scenarios and outcomes on the front end. While some of these tools are already in use, responsible use may be better than a complete prohibition.

Finally, periodically review your policy or guidelines. This is a swift-moving technology, so it is imperative to set a timeline to review and update your guidance based on changes in technology and your individual campus use cases.

More From UE

Using Artificial Intelligence Tools in the HR Lifecycle: Risks to Consider

Additional Resources

University of Pennsylvania: Generative AI — Best Practices and Resources

Yale University: Guidelines for the Use of Generative AI Tools

University of Wisconsin-Madison: Generative AI @ UW-Madison: Use & Policies