Researcher Checklist to Evaluate AI Tool Utilization
Here is a quick-reference checklist designed for researchers to verify their compliance with ethical, legal and institutional standards before submitting a manuscript or grant proposal.
Pre-Submission AI Audit Checklist for Researchers
Authorship and Accountability
[ ] AI is not listed as an author: I have ensured that no generative AI tool (e.g., ChatGPT, Claude) is credited as a co-author on this submission.
[ ] Human verification: I have rigorously fact-checked all AI-generated text, data analysis and citations for accuracy, recognizing that AI tools can "hallucinate" or fabricate information.
[ ] Full responsibility: I acknowledge that I am ultimately accountable for the integrity, originality and validity of the entire submitted work, including any AI-assisted components.
Data Privacy and Security
[ ] No sensitive data exposed: I verify that no Protected Health Information, personally identifiable information or human subject data was entered into unauthorized, public-facing AI tools.
[ ] Protection of intellectual property: I did not upload unpublished research data, proprietary institutional information or confidential collaborator data into public AI platforms.
Transparency and Disclosure
[ ] Explicit disclosure: I have clearly disclosed the use of generative AI in the appropriate section of the manuscript or proposal (e.g., Methodology, Acknowledgments or Cover Letter).
[ ] Detailed documentation: I have maintained a record of the specific AI tools used (including the version/date), the prompts submitted and exactly how the output was used to support the research process.
Funder and Publisher Compliance
[ ] Publisher guidelines: I have reviewed and complied with the specific AI usage policies of the target journal or publisher.
[ ] Funder guidelines: If submitting a grant, I have adhered to the specific generative AI policies of the funding agency.
Peer Review Prohibition (If applicable)
[ ] Confidentiality maintained: If I am acting as a peer reviewer for this or any other cycle, I have not uploaded confidential grant applications, unpublished manuscripts or review materials into a generative AI tool, as doing so is a breach of confidentiality and explicitly prohibited by agencies like the NIH and NSF.