UWM Guidelines for AI-Generated Media and Content

Generative artificial intelligence (AI) tools can support content creation, but they must be used responsibly. These guidelines are meant to help UWM communications and marketing professionals ensure content developed with AI assistance meets university standards for accuracy, accessibility, privacy and brand integrity. 

In addition to reviewing the guidelines on this page, please see AI at UWM.

Guiding Principles 

AI can be a valuable tool for supporting your work. However, it is not error-free, and all AI-generated content requires appropriate human review and oversight before use. Keep the following considerations in mind when using AI tools:

  • Accuracy first: All content must be factual and verified using reliable, non-AI-based sources. 
  • Human oversight required: AI is fallible and should not be the final reviewer or decision-maker. 
  • Protect privacy and data: Sensitive or confidential information must never be shared with AI tools. 
  • Maintain trust and transparency: Content should not mislead audiences about its origin or authenticity. 

Acceptable Uses of AI

AI may be used in limited ways to support content production for marketing and communications. It can help:

  • Brainstorm ideas, outlines or concepts 
  • Draft content, including:
    • Turning talking points into an email to colleagues.
    • Turning bulleted notes into draft copy for a webpage.
    • Converting an outline into step-by-step instructions on how to complete a task.
  • Suggest revisions or improvements to content
  • Rewrite or refine content for clarity and tone 
  • Assist with formatting or structural guidance
  • Adapt content for different audiences or platforms
  • Proofread for grammar and spelling errors
  • Draft text to support accessibility features (e.g., alt text, captions)

All AI-drafted text must be reviewed by a human. In addition, AI should not replace human authorship in official university communications, such as emails from the chancellor and other campus leaders. 

Prohibited Uses of AI

AI may not be used to create or support the development of:

  • Misleading, fabricated or deceptive content
  • Impersonation of individuals or the official voice of the university
  • Fake testimonials, endorsements or quotes
  • Images, audio or video that falsely depict a person saying or doing something they did not actually say or do
  • Misrepresentation of university facilities, programs, experiences or outcomes
  • Photos or video. This includes:
    • AI-generated images or video depicting UWM’s campus or facilities
    • AI-generated depictions of human beings, including UWM students and employees
    • Any image that misrepresents reality 
  • Audio or voice generation. This includes:
    • AI-generated voiceovers or narration
      • Digital accessibility exception: AI-generated narration may be used for accessibility features such as audio descriptions, provided the voice is not intended to represent a real person and is used solely to improve access to university content. When feasible, the use of AI-generated narration should be disclosed.
    • Imitations of real people and/or human voices
    • Filling in “gaps” in human narration
  • Replication or modification of UWM logos, seals or mascot images. Use only approved UWM brand assets.

AI-Assisted Content Production and Review Process

UWM employees remain responsible for the accuracy, quality and integrity of any content created with the assistance of AI. To ensure content meets university standards, follow the process below.

  • Define how AI will be used and ensure its use complies with university and unit guidelines. This includes informing your supervisor of your planned use.
  • Gather accurate, UWM-specific information from reliable and authoritative sources.
    • AI should never be used as a primary source of information because it can generate inaccurate, misleading or fabricated content.
    • Primary sources are original documents or people with direct knowledge of a subject. For example, the senior vice chancellor for Finance and Administrative Affairs is a primary source on the UWM budget, and the university architect is a primary source on a building project.
  • Use AI to support brainstorming, outlining or drafting.
  • Review and revise content to reflect UWM’s voice and editorial standards.
  • Independently verify all facts, statistics, quotations, links and references.
  • Review content for accuracy, bias, clarity, accessibility and compliance with university policies.
  • Obtain any required supervisor, stakeholder or subject matter expert approvals.
  • Complete a final human review before publication or distribution.

Privacy

UWM is legally and ethically obligated to protect individual and institutional data. UWM’s Information Technology Services recommends using Microsoft 365 Copilot Chat because it has enterprise-level data protection applied when you log in through your UWM Microsoft 365 account. The version of Copilot Chat included in UWM’s Microsoft suite is the only version approved for use with UWM data.

Entering moderate- or high-risk data into the free version of Copilot Chat or other generative AI tools is like posting that data on a public website. AI tools collect and store data from users as part of their learning process, and they may share that data with other users outside the university.

Here are a few examples of the types of data that fall into different categories:

  • Public data (low risk)
    • Degree requirements 
    • University approved press releases 
    • Final budget data 
  • Internal data (low risk)
    • Course evaluation summary results for a program or department 
    • Draft, preliminary, and unofficial budget information
    • Funding strings, purchase requisitions and purchase orders 
  • Sensitive data (moderate risk)
    • Graded work, grade book, etc. 
    • Name and preferred name 
    • Date of birth 
    • Directory address and phone number 
  • Restricted data (high risk)
    • Social Security number or partial Social Security number 
    • Driver’s license number 
    • Resident card or similar identification 
    • Credit/debit card number 

Review UW System Administrative Policy Guidance 1031 Information Security: Data Classification for additional guidance.

Bias and Discrimination

Because predictive language models generate text similar to the text they were trained on, they risk perpetuating bias and discrimination in the way they construct language and in their word choices.

All content generated by an AI tool should be reviewed by at least two people — the prompt writer and their supervisor or a subject matter expert — to help control for bias.

It is always recommended that stakeholder review of marketing materials include a group that is diverse in race/ethnicity, class, gender, sexual orientation, country of origin, religion and other characteristics. UWM’s Office of Community Empowerment & Institutional Inclusivity is available as a resource to help facilitate the review of materials.

Translation and Multilingual Content

Some AI tools, including ChatGPT, work in multiple languages. They can be used to translate English text into Spanish and other languages. When this is done, the new text should always be reviewed by a fluent speaker of the target language.

It is important to note that AI tools’ facility varies with different languages. For example, as of September 2026, ChatGPT was capable of generating text in more than 100 languages. However, it provided a disclaimer that “my proficiency varies by language. I’m strongest in English and generally perform best in widely used written languages. For some less common languages or dialects, I may be less accurate.”

Additional Guidelines

  • All AI-assisted content must meet UWM digital accessibility standards. Do not rely solely on AI tools to ensure accessibility compliance.
  • Do not use AI to draft crisis communications, legal content, health and safety information, or communications from campus leadership without prior approval from your supervisor and the MarComm vice chancellor.
  • Follow all applicable copyright, licensing and intellectual property requirements. Do not assume AI-generated content is free to use or distribute; confirm usage rights when necessary.

Social Media Trends and AI-Generated Content Disclosure

In very limited circumstances, UWM may participate in social media trends that involve AI-generated content. Such use requires advance approval in writing from the MarComm vice chancellor and must comply with all university branding, accuracy and authenticity standards. To obtain approval, begin by submitting a proposal via MarComm’s service request form. Select “social media” as the area you need help with and attach your proposal as a Word document.

If approved by the vice chancellor:

  • Clearly disclose the use of AI within the content, in the caption or through platform-provided labeling.
  • Use platform-provided AI content labels when available.
  • Include additional disclosure when necessary to ensure audiences understand that AI was used.