10 Essential Rules for Using AI in Grant Writing: A Researcher's Guide

Artificial intelligence tools are rapidly becoming part of the academic research process, and grant writing is no exception. Understanding how to use them responsibly matters more than ever. Drawing on the PLOS Computational Biology article 'Ten Simple Rules to Leverage Large Language Models for Getting Grants' by Elizabeth Seckel, Brandi Stephens, and Fatima Rodriguez, this guide offers practical, actionable advice for researchers. The timing is especially relevant: the NIH recently released updated guidance on AI use in grant applications (NOT-OD-25-132). Whether you are preparing a proposal or simply staying current with best practices, these ten rules will help you integrate AI thoughtfully and in line with federal expectations.

Funding agencies differ widely in their policies on AI use. Some permit it with disclosure, while others restrict it outright. The NIH, for example, issued a policy on July 17, 2025 (NOT-OD-25-132) that prohibits applications substantially developed with AI. Before using tools such as ChatGPT, review the funder's website for the latest guidance.

Many AI tools may save user inputs to improve future models, which can expose original research ideas to other users. Avoid pasting unpublished data, budget details, or your Specific Aims into public chatbots. Instead, use secure AI tools available through your institution. Locally hosted models such as PrivateGPT or LM Studio can run entirely offline, offering greater privacy though often less power than mainstream tools. Stanford also provides its own secure GPT tool, the AI Playground.

Begin with your own words and ideas. AI can help refine language later, but it should not generate your core content. Reviewers can often detect generic prose that lacks the author's voice. Your grant must reflect you as a scientist: your scientific ideas, preliminary data, and novel approach, described in your own words.

Detailed prompts produce better feedback. Include the grant program and the section you need help with, and ask the AI to check tone, clarity, or grammar rather than rewrite the content. The researchers also found AI performs better on shorter sections. For example, a postdoctoral scholar might ask for help improving the clarity of Career Development Plan training goals while aligning them with the funder's review criteria, focusing on structure and tone rather than wholesale rephrasing.

AI can fabricate sources, citations, and even data. Verify every fact, reference, and claim it generates. If you use AI to brainstorm ideas, fact-check them thoroughly before incorporating anything into your final draft.

AI-generated content may contain plagiarism or bias. Treat it as a source of edit suggestions and brainstorming support rather than a replacement for your writing. Keep your voice and expertise at the center of the proposal.

AI is more than a shortcut; it can make you a stronger grant writer. When you receive a suggestion, ask why it works. Does it improve clarity, strengthen an argument, or make the language more concise? Treating each exchange as a learning opportunity sharpens your own writing instincts over time.

Image-generating tools such as DALL·E 3 (via ChatGPT Pro), Midjourney, and Adobe Firefly can help sketch figure concepts or gather feedback on visuals. Ensure that any images are scientifically accurate and that the funder permits AI-generated figures. An example prompt might request a conceptual timeline of clinical trial phases, from early discovery through implementation, with key milestones such as IRB approval and patient recruitment.

AI has notable limitations in grant writing. Colleagues and mentors can identify issues AI cannot yet detect, such as scientific logic, cultural tone, and strategic fit. The best results come from pairing AI support with human insight.

Test different tools and prompts to discover what works best for you. Ask the same question in different ways, and try prompts on material you already know well. This helps you understand each model's strengths and weaknesses before relying on it for high-stakes writing.

Used responsibly, AI can streamline the grant writing process while preserving the scientific integrity and personal voice that reviewers value. By checking funder policies, protecting your ideas, verifying outputs, and combining AI with human expertise, researchers can integrate these tools in ways that meet evolving federal expectations and strengthen their proposals.