NIH Tightens Rules on AI in Grant Applications
Last Updated on September 3, 2025
September 2025 marks a turning point for researchers using AI in funding applications.
The National Institutes of Health (NIH) has released new guidance aimed at protecting the fairness and originality of its research funding process. This policy goes into effect for applications submitted on or after September 25, 2025—and it carries major implications for anyone using AI tools in proposal development.
Two Big Changes
-
AI-generated text is not acceptable.
-
NIH explicitly states that applications “substantially developed by AI” or containing sections “substantially developed by AI” will not be considered original work.
-
If AI use is detected after funding, cases may be referred to the Office of Research Integrity for possible misconduct. Penalties could include cost disallowance, suspension, or termination of awards.
-
-
Application limits per investigator.
-
Beginning this cycle, NIH will only accept six applications per Principal Investigator (PI) per calendar year (excluding training grants and conference grants).
-
This cap is intended to prevent the flood of AI-assisted submissions that overwhelm review panels—some PIs have recently submitted 40+ applications in a single round.
-
Why This Matters
NIH acknowledges that AI can reduce the administrative burden of proposal preparation, but it emphasizes the risks: plagiarism, fabricated citations, and erosion of originality. For researchers, freelancers, and small businesses supporting NIH applications, this policy reinforces two truths:
-
Human oversight is non-negotiable. AI can assist with formatting, language polish, or idea organization, but the intellectual core must come from the applicant team.
-
Strategic focus is more valuable than volume. With a six-application cap, PIs (and their support teams) must prioritize quality over quantity.
What Freelancers and Agencies Should Do
-
Audit workflows now. If you or your clients are using AI in drafting, document where and how it’s applied.
-
Educate clients. Make sure investigators understand that NIH views substantial AI drafting as non-original.
-
Refocus services. Position yourself as an expert in refining, fact-checking, and compliance—roles where human expertise adds undeniable value.
The Bigger Picture
This is part of a growing trend: federal funders tightening rules around AI use. NIH is also investing in detection technologies, meaning enforcement will only get stricter. For professionals in research support, the safest path is transparent, limited, and well-documented use of AI tools—never a replacement for human originality.