Responsible AI Use In Emory Marketing and Communications

Guidelines for Emory Communicators

Generative artificial intelligence brings great potential to enhance the efficiency and effectiveness of marketing and communications professionals when used in appropriate ways. At the same time, improper use of AI tools can introduce security risks and produce inaccuracies, biased perspectives and other unintended consequences.

The following guidelines are intended to help Emory University communicators make the most of these evolving tools while upholding our commitment to ethical, trustworthy and authentic communications on behalf of the university.

These guidelines apply to Emory University's institutional communications and marketing work. Students, faculty and staff using generative AI for academic, research or other purposes should consult with their instructors, managers or leadership for guidance relevant to their work, in addition to following applicable Emory policies.

Acceptable AI Use

Any use of AI in institutional marketing and communications must follow applicable Emory guidelines and policies.

Within those guidelines, Emory communicators and marketers use AI tools to support our work in ways including (but not limited to):

  • Brainstorming (interview questions, headlines, event marketing ideas, etc.)
  • Preliminary copyediting of human-generated content
  • Revising content to reduce wordcount, ensure appropriate reading level, or change format
  • Analyzing marketing data and media coverage trends
  • Summarizing information
  • Translating complex research descriptions into simpler language
  • Optimizing content for SEO
  • Transcribing audio and video files
  • Drafting early versions of non-bylined, routine web content, social media posts or marketing copy, followed by human refinement
  • Drafting highly structured, impersonal, routine content (e.g. Q&A documents, simple training videos, meta descriptions)
  • Customizing marketing copy for specific audiences
  • Suggesting photo descriptions or alt text for accessibility
  • Creating illustrations (within the parameters listed in the "Special cautions" section below)
  • Routine editing of Emory-owned images or video to optimize them for publication without altering their basic content or meaning. For example, this may include removing distracting background clutter, expanding image backgrounds to a desired size, eliminating brand names from clothing or blurring nametags for privacy, and basic retouching.

All Emory marketing and communications content created or edited with support of AI tools should be carefully reviewed by at least one professional communicator before publication.

Disclosure Of AI Use

Emory communicators are generally not required to publicly disclose AI use in instances outlined in the "acceptable use" section above, as long as the content has gone through careful human review prior to publication.

When providing content that has been generated, influenced or modified by AI to others for review (colleague, supervisor or other "approver", subject matter expert, etc.), communicators should disclose to those individuals how AI was used. This may include disclosing acceptable uses listed in Section 1, when the work product has been significantly shaped using AI.

AI use should be disclosed in writing in the cases listed below:

  • When required by law or vendor policy (eg. YouTube, Instagram). Be sure to check the vendor requirements for attribution.
  • When the generated content cannot be checked by a professional communicator or other knowledgeable individual before posting (eg. chatbot replies)
  • When images or videos are generated by AI
  • When existing images or videos are altered in a way that changes important facts or context. (Also see the "When not to use AI" guidance below.)
  • When generative AI provided a level of contribution that would typically be cited as a source or credited as a co-author.

Credits can be placed under the work using language such as:

  • For copy: "Portions of this story were created or edited using generative AI."
  • For imagery: "This image was created or edited by image-to-image or text-to-image generative AI."

When Not To Use AI

Communicators should avoid using generative AI when working on certain kinds of Emory marketing and communications projects or content.

AI should not be used:

  • In any case that would inappropriately publicize information via putting it into the AI tool. Embargoed, sensitive or legally protected content, including intellectual property, protected personal information (as defined by FERPA and HIPAA, for example), images or videos that are unconsented or not authorized for public distribution, and internal, confidential or restricted data (refer to Policy 5.12, "Data Classification Definitions" section), should not be uploaded to AI tools unless the tools are explicitly approved by Emory for such use. See the Emory Approved Secure AI Technology Registry for more information on tools and related guidance.
  • To generate, fully or in part, news releases or feature articles that are bylined or otherwise presented as original Emory content.
  • To generate, fully or in part, quotes, speeches, or other messages coming from individuals' personal "voices" as an official Emory communication.
  • To generate images for use in Emory marketing or advertising contexts.
  • In any crisis communications that require rapid response to serious or developing situations (eg. mass emails, social media posts or press statements related to campus emergencies or unexpected events).
  • To use a real individual's likeness or voice to generate new, materially altered information (eg. AI-generated or modified photos or videos that do not reflect what that person actually looked like, did or said).
    Exception: The content is
    1. clearly labeled as AI-generated,
    2. cannot be misinterpreted as "real," and
    3. the person has agreed to and approved the final product in writing.

AI should be used with special caution:

  • When creating social media posts, given many platforms' disclosure requirements and the potential for unintended errors to be rapidly and widely shared.
  • When generating novel images or videos for use in editorial or social media contexts. Any such content should only be produced using tools that include some protection against copyright infringement  (for example, Microsoft, OpenAI, Anthropic, Google, and Adobe). Consider setting an expiration date on such content where possible. Note also the potential for AI-generated multimedia content to draw negative feedback due to the environmental impacts associated with its creation.
  • When generating any content using Emory trademarks, such as Emory logos and the images of Swoop and Dooley. Please contact Emory Trademark Licensing in advance for approval and guidance.

These guidelines are effective as of August 2025 and will be reviewed and updated to reflect the evolving use of AI at least annually.