AI Guidelines for Communications and Marketing

Developed by the Office of Communications and the Office of Strategic Brand and Marketing

The Office of Communications and the Office of Strategic Brand and Marketing at FDU provide guidance for all content creators using images, text, video and audio in University marketing, communications and promotional materials.

While AI is already incorporated into many communications tools and presents positive opportunities for our community, it is a developing technology that also comes with unique challenges. Using generative AI for communications on FDU’s behalf requires a firm understanding of its use, cases and limitations, knowledge of the ethical and potential legal considerations involved and, most importantly, human oversight.

The following guidelines have been developed to help FDU communicators leverage generative AI as a tool for effective content generation. Here, the word “content” is used in a broad sense: text, graphics, photography, videos, music, voice etc.

Note: These guidelines do not apply to educational and classroom settings and should not be construed as official FDU academic policy. Because AI technology is advancing quickly, this is a living document that will be updated as best practices evolve.

Our Four Guiding Principles

AI tools are “trained on” existing content, and, in response to user input, synthesize new material. They are statistical models that generate responses based on patterns in their training content.

AI does not understand the concept of “truth” and, as such, has demonstrated that it can have biases and “hallucinate” — i.e., make up facts and pull information from source data without regard for citations or intellectual property rights. It does not engage with moral or ethical considerations, cannot think critically and is not a replacement for human imagination or discernment.

For these reasons, it is critical that FDU communicators adhere to these four core guiding principles:

  1. Human-Centered Activity – Use AI as a Tool
    AI tools are just that: tools. They should assist — not replace — the people who make content. The breadth of your communications should never be turned over to AI, but AI can support your process. AI is no substitute for the creativity, empathy, attention to detail and professional expertise practiced by FDU content creators, but it can be used to enhance our work.
  2. Review Everything
    We are ultimately accountable for the content we create. AI tools are statistical models and cannot be regarded as an authority on any topic. Their output is influenced by existing material, and this material can be biased, without context, out-of-date and incorrect. AI also does not fact-check or reliably take copyright or other intellectual property rights into account.

    We highly recommend you ask for citations in your prompts. Regardless of citations, it is critical to review and edit any outputs produced by AI to eliminate bias, fact-check content and avoid plagiarism.

  3. Protect Data and Privacy
    Some AI tools may store what you input and use it in future output(s), and this information also may not be encrypted. It is critical that we do not input any information that is personal or confidential (yours, FDU’s or anyone else’s), as this may be in violation of University policy as well as state or federal privacy laws such as FERPA (Family Education Rights and Privacy Act).

    Examples of protected information include: Names, demographic information, birthdays, original student ideas, student and/or employee ID numbers.

    Branding restrictions:     Do not input licensed, copyrighted or proprietary content such as official FDU logos or wordmarks into public AI tools.

  4. Be Transparent
    Be forthright when you are using AI-generated content. If you have used AI to meaningfully change an image or other content, consider a citation to that effect, referencing the specific tool you used (e.g., ChatGPT, OpenAI, etc.).

    Sample statements are as follows:
    “During the preparation of this work, the author(s) used [AI Tool Name and Version] to [describe task], then reviewed and edited the content to take full responsibility.”
    “The author(s) used [AI Tool Name and Version] to improve readability and, after careful review, takes full responsibility for the final content.”
    “Image generated using [AI Tool Name, Company Name, Date].”

    While the legal landscape related to AI use in advertising and marketing continues to evolve, certain state laws require disclosure of AI use and may lead to fines if disclosure is not present. Beyond the United States, as applicable, guidelines such as those defined by the EU Artificial Intelligence Act, et al., must be adhered to.

Imagery and Creative Tools

Core Imagery Guideline

AI-generated imagery is not permitted in any University marketing, communications or promotional materials, including on our websites and other communication platforms. This is to honor our audience’s expectations that our content, often featuring our people, places and experiences at the University, is authentic.

Acceptable AI Tools in the Creative Process

It is acceptable to use AI tools as helpers in creative processes, restricted to the following areas:

  • Audio editing: General clean-up, but not to include the alteration of a person’s spoken words.
  • Sound design: The creation or editing of background, ambient or environmental sounds.
  • Content-aware photo expansion: Expanding what is already in a scene, provided it does not invent entirely new contextual content.
  • Design layouts: Assisting in creating initial graphic layouts or grids.
  • Object removal: Removing distracting objects from photography, provided that removal does not fundamentally alter the meaning or context of human subjects or their environment.
  • Administrative workflow: Transcription, note-taking, workflow optimization, office management and basic research.

Exceptions

  • AI may be used as a tool to create conceptual or illustrative content rather than representative content (e.g., an image of an amoeba, or a nonrepresentational background design).
  • Exceptions must be approved by the Office of Communications and should be used only when original images or graphics are not available. Any content, messaging or assets created and intended for external consumer marketing must be approved by the Office of Strategic Brand and Marketing. Examples of external consumer marketing are advertising (print and digital) and promotional collateral (e.g., flyers, mailers, banners, posters, etc.).
  • If approved, AI-generated or stock photography used for final content must be clearly labeled. Sample wording such as “Image generated using [AI Tool Name, Company Name, Date]” may be used.

When to Use AI: Detailed Use Cases

AI tools can be especially helpful with tasks like creative brainstorming, taking notes, refining ideas, generating code, formatting text and editing images. It can also help integrate information and take first passes at editing content. Still, AI tools should never be used as the final source for facts or editing. Instead, they should only be used to help save time on routine tasks and provide a jumping-off point for creative ideation.

Examples of Acceptable Use

  • Brainstorming or creating mood boards: AI tools can create first-draft images or compile pictures and images to use as internal inspiration.
  • Research and synthesis: AI tools can create summaries for quick overviews of new topics. This research must be verified by humans and is not a substitute for scholarly research.
  • Display Information: AI tools can generate charts, graphs and depictions of data or information.
  • Editorial help: You can have AI tools take a first pass at editing your content to match AP Style, etc. But be aware that it may not be using the most up-to-date versions of these manuals and your content will still need to be read by a person. AI tools can also make suggestions for shortening your content, looking for repeated words or ideas or suggesting more efficient language. All final editorial reviews should be conducted with human oversight before publication.
  • Note-taking and transcription: AI tools can help you take notes and transcribe interviews or meetings to save time. The results are imperfect and should not be assumed to be a verbatim transcript. AI-generated transcripts should be carefully reviewed before they are shared for anything more than your personal use and reference.
  • Editing images: AI can help you augment or enhance an image, like retouching an existing photo, cleaning up a grainy shot or expanding an image. People, places and other characteristic elements that tell the FDU story should not be generated by AI. Adhere to existing best practices for photo editing and stay true to the goal of presenting factual information.
  • Creating abstract imagery: AI can be used to create images (not photographs) that are not intended to be interpreted as portraying reality. After you have created the image, use a web image search tool to make sure it has not been plagiarized from another source. Be transparent and credit the image as AI-generated.

Examples of Unacceptable Use

  • Creating photographs or photorealistic images: Content should not use photographs or photorealistic images that have been created entirely by AI, especially if they have human subjects or identifiable FDU students, staff, faculty or campus locations. AI can be used to edit real images, as defined in the Core Imagery Guideline section above, but not to create fictional ones.
  • Captions: AI-generated captions may not be accurate or meet our strict digital accessibility requirements.
  • Sensitive, urgent or private content: Communications that are sensitive, urgent or require protected or personal information should not be created using AI tools. They should be created by professionals who are authorities on the subject matter, aware of the clarity and empathy these communications require and mindful of protecting personal information.
  • Authored content: Any content with an official byline should be the original work of the author.

Additional Considerations: Photography, Video and Audio

Exercise extreme caution when using AI tools in multimedia. For example, using an AI voice-over tool for an internal staff training video is acceptable, but using AI to generate the voices of any FDU community member or guest — be it a specific individual or a generic representative — is misleading. If an AI script or narration is used, it should be reviewed carefully and openly disclosed.

AI should never be used to create sound bites or quotes by people who speak in videos. Inputting a real person’s voice into an AI tool endangers that person’s privacy and can be a severe security risk. This is because the recorded voice of a real person can be used to generate unlimited audio files that recreate the sound of the actual person’s voice saying anything, completely detached from what the person said in the recording.

AI-generated music may be appropriate in videos but can very easily infringe on copyrighted material, even if the AI tool uses only a portion of the material. The same is true of photography and art. Consider other human-generated resources first, including licensed, royalty-free music platforms.

Always carefully review AI-generated work. Since AI tools pull from numerous internet sources, consider whether the AI-generated work is reasonably similar to existing copyrighted work.

Responsible AI Usage and Governance

AI technologies can be used to streamline processes, activities and tasks. At the same time, they introduce security risks for staff members and the University itself. We must be mindful of the data we offer to AI platforms and do our part to protect the work of scholars and artists connected to FDU and the larger internet community.

The Office of Communications values freedom of expression while striving to protect privacy and provide responsible work attribution. Although approved usage of AI is encouraged, it should never replace our human-centered approach.

We advocate for responsible AI usage through critical self-governance and accountability. Continuous draft content generated using AI, such as scripts or story drafts, must be carefully reviewed, edited and fact-checked by the Office of Communications to ensure ethical, factual, compliance and brand guidelines are strictly followed.

Final Review Checklist

Before you publish any piece of content that utilized generative AI in its workflow, ask yourself:

  1. Is it accurate? Have all names, facts, dates and historical contexts been verified by a human?
  2. Is it safe? Did I ensure no private student data or copyrighted branding elements were leaked into a public model?
  3. Is it authentic? Does this project reflect the genuine spirit of FDU or does it feel artificially manufactured?

Need Help?

If you are unsure whether AI is appropriate for your project, contact the Office of Communications before publishing. (Hector Torres, University Graphic Designer, at hector@fdu.edu)

You can also download a copy of the FDU AI Guide for Communications and Marketing.