AI Generated Media
Plain and practical for media.
These guidelines were reviewed and approved by the MSU Office of Public Affairs. They will be regularly reviewed and updated to reflect advancements in generative AI technologies and best practices.
They cover the communications and media MSU produces: images, video, audio, and the written work that goes with them. Select any guideline to see what it means in practice.
The guidelines
Review and edit before publishing AI output is a draft, not a finished deliverable.
Treat AI output as a starting point, not a final product. Edit it for accuracy, brand alignment, and tone, and have a person review it before anything is published. Unedited AI content should not go out as a finished MSU deliverable.
Representing real people, places, and events Don't present synthetic media as documentary.
Never use AI-generated material to depict real MSU people, places, or events as though it were an authentic photograph, recording, or account. Audiences should never be led to believe that a fabricated image or scene actually happened.
Altering photographs and video Keep the original context and meaning intact.
Don't use AI tools to change the context or meaning of a photograph or video. Routine adjustments such as cropping, color, and exposure remain fine. Manipulations that misrepresent what actually occurred are not.
MSU brand and trademarks AI should enhance the brand, not alter it.
AI should enhance MSU's brand storytelling and visual identity, not alter it. Don't use AI tools to recreate, imitate, or modify MSU's marks, including the “M State” logo, the university wordmarks, and Bully the Bulldog. Official marks come from the Office of Public Affairs.
Disclosing AI-generated content Disclosure is the default at MSU.
Clearly disclose AI-generated content when it is published. Disclosure is the default at MSU. When AI substantially generated or altered an image, video, or piece of writing, say so.
Likeness and voice Get explicit consent before cloning anyone.
Don't generate or clone a real person's voice or likeness with AI without their explicit consent. This applies equally to MSU faculty, staff, students, and members of the community.
Accessibility content Verify AI captions, alt-text, and transcripts.
AI can speed up alt-text, captions, and transcriptions, but it makes mistakes. Always review AI-generated accessibility content for accuracy before it is published. Do not rely on it unreviewed.
Privacy & security
Protecting sensitive institutional and personal information from exposure to external systems, and complying with all federal, state, and local law, is critical to media work. Never input confidential or sensitive MSU data into generative AI tools.
How to classify and handle university data is already covered in the main AI Guidance. Before putting any MSU information into a tool, review the data classifications and risk matrix there.
Check the data first: see the data classifications and risk matrix on the main AI Guidance page before you paste anything into a tool.
Questions
Questions about these guidelines or about a specific project? Contact the Office of Public Affairs at chris.bryant@opa.msstate.edu.