Offices & Services
Office of Communications

AI Marketing Guidelines

Generative AI is a type of artificial intelligence that can learn from and mimic large amounts of data to create content such as text, images, music, videos, code, and more, based on inputs or prompts. Harvard University supports responsible experimentation with Generative AI tools, but there are important considerations to keep in mind when using these tools, including information security and data privacy, compliance, copyright, and academic integrity. These are defined online at: https://huit.harvard.edu/ai/guidelines. Additional AI guidelines are provided by the Office of the Provost.

AI Use in Marketing and Communications

Artificial Intelligence (AI) is having a significant impact on the marketing discipline. These guidelines are intended to set boundaries on the sanctioned uses of AI in our marketing and communications activities.

Generative AI is a tool, not a replacement for human expertise and judgment. You are responsible for ensuring any content you create using generative AI is accurate and meets quality standards. Always carefully review any AI-generated content before publication.

Prioritize data privacy and security in all generative AI-related work. You should not enter private or proprietary data into generative AI tools that lack data privacy protections.

When using generative AI, prioritize creating outputs that are accessible and inclusive. For additional information regarding digital accessibility guidelines, visit https://huit.harvard.edu/ai/guidelines.

Some Acceptable Uses of AI in Marketing & Communications

Content

  • Creating drafts of written content for blogs, social media, emails, and ads if they are carefully proofread for accuracy and quality by humans.
  • Drafting marketing plans and creative briefs.
  • Developing draft video/audio scripts
  • Creating YouTube video descriptions 
  • Generating transcripts for audio and video content if reviewed for accuracy and quality.
  • Developing blog content (with attribution)
  • Developing descriptive web content

Multimedia (with attribution)

  • Removing privacy-related or undesirable elements from photographs while preserving the substance/intent of the original photograph
  • Creating illustrations or animations for stories and articles as a substitute for stock photography
  • Modifying or enlarging the backgrounds of photographs (while preserving the substance/intent of the original photograph)
  • Developing motion graphics of non-human things, environments, processes as a substitute for stock photography
  • Developing podcasts

Prohibited Uses of Generative AI (examples)

  • Replicating or imitating a person’s likeness or voice without their written consent.
  • Creating headshots or portraits of individuals.
  • Generating images of fictional people without clear AI attribution.
  • Modifying images and video in ways that alter the original intent of the image. 

AI Citation Guidelines

To ensure the authenticity of the content we produce, it is important to be transparent about the use of AI. AI-generated images (e.g. illustrations, data visualizations) must be identified as such by placing the following tag in the lower right corner of the image: Created using AI. Images that are modified, edited, or enhanced (e.g. sharpened, color-corrected) do not require a tag. Content that was primarily developed by AI requires attribution. Content that was inspired by, guided by, or edited with AI does not require attribution. 

Text Attribution

Please follow the APA Style for attribution:

Bibliography Format: Author. (Date). Name of tool (Version of tool) [Large language model]. URL

Bibliography Example: OpenAI. (2025). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/chat

In-text citation example: (OpenAI, 2025)

Image Attribution

When citing AI-generated images in online content, include attribution directly below the image or in the image caption.

Example: (Image generated using Dall-E)

Podcast Attribution

Cite your podcast episode using APA Style. List the AI tool as author, followed by the label “(Host),” the date, the episode title and number, the description “[Audio podcast episode],” the name of the podcast, the production company, and a URL if available. 

Social Media Attribution

On social media platforms, include attribution embedded in the image and in the image caption or post text. 

Example: "AI-generated image of a cosmic landscape using Dall-E."

Prompt engineering for consistent brand voice and tone

Using a consistent brand voice and tone in external communications is important for maintaining a consistent and unified voice. As AI generated content becomes part of the content development process, it’s important to approach this process in a consistent manner as well. 

Here are some best practices for AI prompt engineering:

  1. Be Clear and Specific: Clearly state the topic, purpose, and desired outcome to guide the AI response accurately.
  2. Give IA a persona & skills: To support the goal of being clear, specify a role or persona and set of skills for the AI that aligns with the desired output style or expertise. 
  3. Include Context: Give background information, the intended audience, or a situation for the AI to consider.
  4. Define the Tone and Style: Specify the tone to ensure consistency with our brand 
  5. Use Simple, Direct Language: Avoid complex or ambiguous language in prompts; clarity yields more relevant responses.
  6. Guide the Response Length and Format: Indicate if the response should be brief, detailed, in bullet points, or a summary to control length and format.

Our Voice & Tone:

  • Inspiring and Visionary: Our language should reflect a forward-thinking mindset, emphasizing the transformative impact of science and engineering on society. Use aspirational, vivid language that resonates with curiosity and the ambition to shape the future.
  • Inclusive and Collaborative: Harvard SEAS values diversity and the collaborative spirit of discovery; our language should reflect openness and respect for all perspectives. Choose inclusive language and frame messages in a way that invites a broad audience.
  • Authoritative yet Accessible: Our tone should convey expertise and knowledge while remaining approachable and clear. Use precise, informed language, and explain complex ideas simply when possible.
  • Tone: The tone should be warm, professional, and inclusive. While we maintain a sense of authority, it’s essential that the audience feels welcomed and respected. This tone reflects our commitment to excellence and openness.