Can I Use ChatGPT for Grant Writing?

Can I use ChatGPT for grant writing? Under the AI policies published by NIH, NSF, DOE, NEH and NASA's Science Mission Directorate, yes, within limits that change from one funder to the next. NIH restricts how much of an application AI may produce, NASA's directorate requires a citation, and funders such as Wellcome and the Spencer Foundation want AI use beyond basic editing declared when you apply.

Two checks come before the first prompt. The first is what your funder's policy allows. The second is which kind of ChatGPT account you are signed in to, because OpenAI treats its services for individuals and its business products differently when it comes to training on your content (OpenAI Help Center, read September 2026).

Quick Answer: Can I Use ChatGPT for Grant Writing?

ChatGPT is allowed in grant writing under the published policies of NIH, NSF, DOE, NEH and NASA's Science Mission Directorate, within limits: NIH restricts AI-developed content without banning it, NASA's directorate requires a citation, DOE's standard notice and NEH require disclosure, and NSF encourages it. Keep confidential data out of individual ChatGPT accounts and verify every fact.

What Funders Allow When You Use ChatGPT

Funder AI policies from NIH to the Spencer Foundation allow ChatGPT as an assistant and draw their line at authorship: how much of the application the tool writes, and whether you say so. The line sits in a different place for each funder, so name the funder before you decide anything.

Three policies show where the line falls. NIH says it "will not consider applications that are either substantially developed by AI, or contain sections substantially developed by AI, to be original ideas of applicants" (NIH notice NOT-OD-25-132, released July 17, 2025). UKRI states that "Applicants must not use generative AI tools to generate an entire application, or sections of an application, without human involvement" (UKRI generative AI policy, last updated July 6, 2026). The Spencer Foundation says applicants "may not submit verbatim drafts of content generated by AI" (Spencer Foundation AI policy, listed August 21, 2025).

Foundation statements differ widely. They run from the Spencer Foundation's required disclosure checkbox to a community foundation that asks about AI use and says the answer "will not have any weight on the decision of the committee" (Community Foundation of Henderson County, AI Usage Statement, modified May 20, 2026).

Where a funder's guidelines say nothing about AI, ask before you rely on the tool. Our guide to contacting a program officer before you apply covers how to raise the question without wasting the officer's time.

What NIH's AI Policy Means for Applicants

NIH restricts AI in applications without banning it. Its notice says "AI tools may be appropriate to assist in application preparation for limited aspects or in specific circumstances, but researchers should be aware that using AI comes with its own risks" (NIH notice NOT-OD-25-132, released July 17, 2025).

The policy "is effective for applications submitted to the September 25, 2025, receipt date and beyond" (NIH notice NOT-OD-25-132, released July 17, 2025). As read in September 2026, NIH's list of policy notices shows it as the latest notice on AI in applications, with no later entry revising it (NIH, Notices of NIH Policy Changes, read September 2026).

The risks are written into the notice. NIH says it "will continue to employ the latest technology in detection of AI-generated content to identify AI generated applications" (NIH notice NOT-OD-25-132, released July 17, 2025). If AI is detected after an award, "NIH may refer the matter to the Office of Research Integrity," and it can also act through measures "including but not limited to disallowing costs, withholding future awards, wholly or in part suspending the grant, and possible termination" (NIH notice NOT-OD-25-132, released July 17, 2025).

The same notice caps volume. NIH "will only accept six new, renewal, resubmission, or revision applications from an individual Principal Investigator/Program Director or Multiple Principal Investigator for all council rounds in a calendar year," and the limit excludes T activity codes and R13 conference grant applications (NIH notice NOT-OD-25-132, released July 17, 2025).

For an applicant, the safe reading is plain. Write the aims, the approach and the argument yourself, and keep ChatGPT to limited tasks such as tightening prose you drafted. Keep a note of what the tool did, in case anyone asks later.

A different NIH rule covers the people who score applications. NOT-OD-23-149 "prohibits NIH scientific peer reviewers from using natural language processors, large language models, or other generative Artificial Intelligence (AI) technologies for analyzing and formulating peer review critiques for grant applications" (NIH notice NOT-OD-23-149, released June 23, 2023). That rule binds reviewers, while applicants follow the separate application notice (NIH notice NOT-OD-25-132, released July 17, 2025).

Do You Have to Disclose ChatGPT Use in a Grant Application?

Disclosure depends on the funder: DOE's standard notice, NEH and NASA's Science Mission Directorate require it, NSF encourages it, and NIH's notice sets no disclosure format. Name the funder, then follow its format exactly.

Sources: NASA SARA FAQs, question 32 (page updated October 13, 2026); DOE Financial Assistance NOFO Part 2, Version 4.0 (the copy attached to notice DE-FOA-0003627, July 2026); NEH AI policy (October 23, 2024); NSF notice (December 14, 2023); Wellcome (April 8, 2026); UKRI policy (July 6, 2026); Spencer Foundation policy (listed August 21, 2025); Wenner-Gren Foundation policy (March 19, 2025).

Some funders state that disclosure will not count against you. UKRI says the information "will not affect the assessment process" (UKRI generative AI policy, last updated July 6, 2026). The Spencer Foundation says applicants who disclose "will not be penalized in the review process for doing so, unless the content generated is used inappropriately, produces inaccuracies, or otherwise negatively impacts the proposal" (Spencer Foundation AI policy, listed August 21, 2025).

One more source matters for research applicants. HHS's Office of Research Integrity, in nonbinding guidance on the Public Health Service research misconduct rules, says "Researchers should disclose any generative AI tools used during research and manuscript and grant preparation," because "Disclosure can protect against research misconduct allegations and support reproducibility" (HHS ORI generative AI guidance, August 2026). ORI also notes that "disclosing generative AI tool use does not necessarily preclude a research misconduct finding" (HHS ORI generative AI guidance, August 2026).

Log your use as you go. NASA's four items include the version number and the date and time the tool generated proposal text, and neither is easy to rebuild from memory weeks later.

Where ChatGPT Helps in a Grant Proposal

ChatGPT helps most with text and structure you already own, and least with facts it would have to supply. Funders' own lists of minor uses point the same way: UKRI does not require disclosure of "minimal use" such as translation, improving English, formatting or reducing word count (UKRI generative AI policy, last updated July 6, 2026), and the Spencer Foundation does not require disclosure of AI used "to enhance the grammar, spelling, and punctuation" (Spencer Foundation AI policy, listed August 21, 2025).

The facts are where the risk sits. GrantStation, a grant-database vendor, warns that AI "can invent statistics, mirror generic 'AI-sounding' phrasing, and, if you're not careful, it can also be fed information it shouldn't see" (GrantStation, Grantwriting with AI Starter Kit, read September 2026). NSF puts the burden on you: "Proposers are responsible for the accuracy and authenticity of their proposal submission in consideration for merit review, including content developed with the assistance of generative AI tools" (NSF notice, December 14, 2023).

Citations are a narrower case. ORI's guidance says references in grant applications "are not generally data or results," so an invented citation "may not constitute fabricated data or results" where the references do not serve as data (HHS ORI generative AI guidance, August 2026). That is a legal distinction, not a reason to trust generated references. Check each one against its source before it goes in.

A statement of need should rest on figures you can source, whatever tool tidies the prose. The same rule holds for every other section covered in our guide to writing a grant proposal.

Keep Confidential Material Out of Personal ChatGPT Accounts

Assume that what you paste into a personal ChatGPT account can be used for training until you turn that off. OpenAI states: "When you use our services for individuals such as ChatGPT and Codex, we may use your content to train our models" (OpenAI Help Center, read September 2026).

To stop training on a signed-in personal ChatGPT account on the web, OpenAI's steps are profile icon, Settings, Data Controls, then turn off "Improve the model for everyone," and the setting "applies to your entire account" on every device (OpenAI Data Controls FAQ, read September 2026). A temporary chat "is not used to improve OpenAI models while it remains temporary," though "OpenAI may keep a copy of a temporary chat for up to 30 days for safety purposes" (OpenAI Help Center, Temporary chat, read September 2026).

Business accounts start from the opposite default. OpenAI says: "By default, we do not train on any inputs or outputs from our products for business users, including ChatGPT Business, ChatGPT Enterprise, and the API" (OpenAI Help Center, read September 2026).

Funders add reasons of their own. NASA's Science Mission Directorate warns proposers that anything given to a generative tool as a prompt "will often be stored and may be used by others," and that "this information may aid a potential competitor" (NASA SARA FAQs, question 32, page updated October 13, 2026). UKRI says "Sensitive or personal data of others must never be input into a generative AI tool without formal consent from the individual" (UKRI generative AI policy, last updated July 6, 2026). Wellcome goes further, requiring applicants to "protect confidential information and personal data by not inputting it into generative AI tools" (Wellcome, April 8, 2026).

One new freelance grant writer on Reddit described the same hesitation: "I've been anxious about uploading documents to Claude or AI because some specific info in grants are not exactly open to public" (Reddit thread 1ubd86x, read September 2026). The fix is mechanical. Before you paste, strip names, addresses, salaries and partner details, and keep unpublished results out of any account that can train on them.

Write the Final Draft in Your Own Words

A funder that reads every application says over-reliance on AI shows, so the text you submit should read as your organization wrote it, with your facts and your local detail in every section.

Halifax Foundation NI, a Northern Ireland funder, says "We read every single funding application we receive" and "it is often noticeable when AI has been over-relied on," adding: "These applications are often generic, and do not bring out your own voice or uniqueness" (Halifax Foundation NI, AI position statement, page last updated July 2025). The London Community Foundation in the UK tells applicants "We're not expecting perfect language, grammar, and style" and asks that AI-assisted answers include "sufficient depth, relevant local context" (The London Community Foundation, Guidance on artificial intelligence, updated October 2025).

A workable routine: let ChatGPT tidy a paragraph, then rewrite it from your own notes and add the local fact, the program figure or the partner name that only your organization knows. If a sentence could sit in any applicant's proposal, cut it or make it specific. Before you submit, read the draft against a list of common grant proposal mistakes.

Drafting inside grant software does not change these rules. GrantCue's AI Grant Writing Workspace offers "Section-by-section AI drafting with grant context," "Three refinement modes: refine, expand, or condense," "Funder alignment analysis and scoring," and a "Reusable content library, with answers imported from your past proposals," with "One-click export to Word or PDF" (grantcue.com, October 2026). Starting from your own stored answers keeps your organization's wording in the draft, and anything the AI writes still counts as AI use under the funder's rules. The GrantCue features page notes that "Available features depend on your plan" (grantcue.com, October 2026).

FAQ

Can I Use ChatGPT to Write a Grant?

Under NIH's and UKRI's policies, yes as a writing assistant and no as the author of the whole application. UKRI says applicants "must not use generative AI tools to generate an entire application, or sections of an application, without human involvement" (UKRI policy, last updated July 6, 2026), and NIH will not treat applications substantially developed by AI as the applicants' original ideas (NIH notice NOT-OD-25-132, released July 17, 2025). Draft with it, then rewrite in your own words.

Can I Use ChatGPT for an NSF Proposal?

Yes, with disclosure encouraged. NSF encourages proposers "to indicate in the project description the extent to which, if any, generative AI technology was used and how it was used to develop their proposal" (NSF notice, December 14, 2023). Its policy supplement effective December 8, 2025 also counts fabrication, falsification or plagiarism committed through AI-based tools as research misconduct (NSF PAPPG 24-1 Supplement 1, December 8, 2025), so check every claim the tool touches.

Can I Use ChatGPT for a Foundation Grant?

Check the foundation's own guidance, since the rules vary. The Spencer Foundation requires a disclosure checkbox and bars verbatim AI drafts (Spencer Foundation AI policy, listed August 21, 2025). The Wenner-Gren Foundation strongly encourages disclosure in a confidential section not visible to reviewers (Wenner-Gren Foundation AI policy, March 19, 2025). The Community Foundation of Henderson County says its AI-use question "will not have any weight on the decision of the committee" (AI Usage Statement, modified May 20, 2026). If the guidelines are silent, ask.

Is It Cheating to Use ChatGPT for a Grant Proposal?

Not when the funder allows it and you follow its terms. The Spencer Foundation's policy draws the line in plain words: generative AI "may be used as an assistive tool" and not as "a substitute author" (Spencer Foundation AI policy, listed August 21, 2025). Use becomes a problem when it breaks the funder's rule, such as undisclosed use where disclosure is required, or an application the tool wrote where the funder requires your authorship.

Will the Funder Know I Used ChatGPT?

NIH says it looks: it "will continue to employ the latest technology in detection of AI-generated content to identify AI generated applications" (NIH notice NOT-OD-25-132, released July 17, 2025). Detection is unreliable, though. OpenAI withdrew its own AI classifier "due to its low rate of accuracy" (OpenAI classifier announcement, retirement note dated July 20, 2023). The dependable approach is to disclose where the funder asks and to submit text you would defend line by line.

Before Your First Prompt, Find the Funder's AI Clause

Open the funding announcement and the funder's application guide, and search both for the terms artificial intelligence and generative AI. Write down whether AI use is limited, whether disclosure is required and where it goes, and whether the funder restricts what may go into outside tools.

Then set up the account. On a personal ChatGPT account, turn off "Improve the model for everyone" first; if your organization has ChatGPT Business, use that instead. Keep a short log of each session with the tool, its version, the date and time, and what it did, which covers the items NASA's Science Mission Directorate asks proposers to cite (NASA SARA FAQs, question 32, page updated October 13, 2026).

When you are ready for the next opportunity, browse open grant announcements and read each one's AI clause before the first draft.