The National Institutes of Health announced it will reject applications for research funding if it detects substantial sections are generated with artificial intelligence and could seek post-award penalties if it later determines AI was used in funding applications.
Additionally, NIH said researchers may only submit up to six new, renewal, resubmission or revisions applications in a calendar year. The limit does not apply to proposals for NIH training grants (“T” activity codes) or conference grant applications (“R13”).
The new policy – outlined in NIH guidance NOT-OD-25-132 – goes into effect Sept. 25.
The Division of Research is sharing the guidance with the Clemson research community for awareness. Questions can be submitted online here. Clemson Computing and Information Technology (CCIT) and University Compliance and Ethics have provided helpful guidelines on the use of AI here.
In its guidance, NIH said it has noticed instances of principal investigators submitting large numbers of funding applications, some generated with AI support, that are unfairly straining its proposal-review processes.
The agency also noted that the use of AI in proposals could result in unintended plagiarism, fabricated citations or other kinds of research misconduct. NIH said it will use technology to detect AI and, if detected after projects are awarded, could seek enforcement actions, including but not limited to disallowing costs, withholding future awards or suspending or terminating the grant.
