When & How to Disclose AI Use

Disclosure protocols should be tailored to the specific context in which the work is produced and disseminated.  Always verify institutional, departmental, journal, or platform-specific policies for disclosure of AI use before publishing or submitting content. Disclosures must be proportionate and specific, reflecting the actual role of AI in the work.

Please consider the following practices to help ensure transparency, accountability, and compliance with ethical standards and legal requirements:

AI use does not always need to be publicly disclosed when the final version has been reviewed and approved by a human prior to publication.

AI use should be disclosed in writing when  any of the following apply:  

Legal, Policy, or Platform Requirements 

• Disclosure is required by law, vendor policy, or platform rules (e.g., YouTube, Instagram).
• Action: Check and comply with each platform’s specific attribution rules.

Lack of Human Oversight 

• The content cannot be reviewed by a knowledgeable human professional communicator or subject-matter expert before publication (e.g., chatbot replies, dynamically generated output, automated outputs).

AI-Created or AI-Modified Media 

• Images or videos were generated using AI tools.
• Existing visuals were altered by AI in a way that changes facts, meaning, or context.

Substantive AI Contribution 

• AI contributed in a meaningful or creative way—for instance:
• Generating or organizing core ideas or hypotheses
• Drafting or summarizing major sections of text
• Conducting or structuring an analysis
• In these cases, AI acted beyond a supportive tool and should be acknowledged.

Unless required in specific contexts, disclosure may not be needed when AI is used only for minor or stylistic assistance, such as: 

• Grammar, spelling, or formatting corrections
• Text cleanup or readability improvements
• Routine data visualization or layout adjustments

When disclosure is required, it should be clear, brief, and placed directly with the content. 

When disclosure is warranted, the method and placement should be appropriate to the audience and context:  

• Public-facing materials (e.g., websites, publications, patient communications) include a conspicuous and visible footnote, acknowledgment, or statement. 
• Internal communications: apply discretion, though transparency is encouraged. 

When applicable, cite AI-generated content according to the relevant formal style guide (APA, MLA, Chicago, or AP Style).  

Suggested Disclosure Language

At Emory, we believe in the power of technology and are committed to using artificial intelligence (AI) responsibly. Please note that [name of AI tool] was used to assist in generating [brief description of content]. While AI contributed to this content, it has been reviewed and approved by [name or role of the responsible individual or office].  

When AI is used to create multimedia content (e.g., digital portraits or synthetic audio), a clear disclaimer should accompany the content. This disclaimer should identify the tool used and describe its role in the creative process. 

In academic, professional, or publication contexts, all substantive AI involvement in the creation or modification of multimedia should be explicitly documented to ensure transparency and uphold integrity.   

The Office of General Counsel, in coordination with the Office of Compliance, should be consulted prior to using any AI-generated or AI-modified images that may carry clinical implications, including publication of altered patient care areas for marketing purposes.

Users are required to comply with their departmental or unit guidelines, which should be consistent with Emory RAI's principles and guidelines.