Yes, you can use AI on your grant. Here's how to do it without getting caught out.

The short answer: yes, you can usually use AI to help write a grant application in Australia. Funders that have published a position, such as the Australian Research Council, NHMRC and the City of Melbourne, allow AI as a support tool, but you remain responsible for every word you submit. Check your program's guidelines for any AI rules, keep confidential details out of public AI tools, and check every fact before you sign the declaration.

If you've been wondering whether using AI on your grant is cheating, or whether an assessor will mark you down for it, you're not alone. Most funders haven't said much, and the ones that have are fairly practical about it. Here's what they say, where AI helps, where it gets people into trouble and how to use it safely.

1. Is it allowed? What Australian funders say

There's no single national rule. Each funder sets its own position, and many programs don't mention AI at all. Here are three published examples:

• Australian Research Council (ARC). Its 2026 policy on generative AI recognises that applicants may use AI to test ideas, improve language and summarise text. It asks applicants to take care that AI doesn't introduce flaws or incorrect data.
• NHMRC. Its revised policy, dated 28 April 2026, lists use by applicants in preparing grant applications as appropriate, as long as a human reviews all outputs and stays accountable.
• City of Melbourne. The Community Inclusion Grants 2027 application form says AI tools can be used to help research and write your application, then lists the risks and what you should do if you use them.

The ARC and NHMRC fund research, and City of Melbourne is one council, so your program may take a different view. If your guidelines or application form say nothing about AI, the usual rules still apply: the content must be true, accurate and yours to submit. If you're unsure, ask the program contact.

2. You're still the author

This is the point every published policy agrees on. Using AI doesn't move responsibility anywhere else.

• The ARC says the applicant is "considered to be the author" of the content they submit.
• NHMRC says applicants must certify that their information is accurate, and are accountable for factual errors, including those that come from using AI.
• City of Melbourne says you're responsible for everything in your application, including anything generated or supported by AI.

Most applications also end with a declaration: a formal statement that the information you've provided is true and accurate, usually signed by an authorised person such as a CEO, secretary or president. The Victorian Government's grant terminology guide describes it this way. When you sign it, you're vouching for every sentence, whoever or whatever drafted it.

3. Where AI helps

Used well, AI can take a lot of the grind out of an application. Common, low-risk uses include:

• Structure. Turning a criterion into a list of the parts you need to answer, or suggesting an order for your points.
• First drafts. Getting words on the page from your own notes, so you're editing rather than staring at a blank box.
• Editing to word limits. Cutting a 320-word answer down to a 200-word limit while keeping the specifics.
• Plain English. Tidying grammar, shortening long sentences and removing jargon. NHMRC notes that AI can help reduce language barriers.
• Checking against the criteria. Asking whether each answer covers every part of the question it sits under, then deciding for yourself.

In each case, the facts come from you. AI is shaping how you say it, not deciding what's true.

4. Where it goes wrong

The problems tend to be the same ones, again and again:

• Invented facts. AI tools can produce confident statements that aren't true: a statistic, a partnership, a past result, a number of participants. NHMRC warns that AI can produce "convincing but inaccurate content". If it ends up in your application, it's your error.
• Generic answers. Without your real details, AI writes answers that could describe any organisation. Assessors can't score "this project will greatly benefit the community", however polished it sounds.
• Sounding like everyone else. When many applicants use public AI chat tools with similar prompts, answers can start to read alike. City of Melbourne asks applicants to make sure the application reflects their own knowledge, experience and voice, and to avoid relying heavily on large sections of AI text.
• Outdated or wrong program details. A general AI tool may not know the current round's rules, dates or caps. Always work from the guidelines for the round you're applying to.

GrantAlly drafts answers from the project details you give it, not from guesses, and flags statements your information doesn't support so you can confirm them, add evidence or remove them. You review every answer and submit it yourself. See how GrantAlly uses AI responsibly.

5. Keep confidential details out of public tools

This is the risk people think about least. When you paste text into a public AI chat tool, you may lose control of where it goes. NHMRC's policy advises you to assume anything entered into an AI tool could be made public, and to use only the minimum information you need.

Be careful with:

• personal information about clients, participants, staff or volunteers
• commercially sensitive figures, pricing, quotes or supplier details
• unpublished ideas, designs or intellectual property
• anything a partner has shared with you in confidence.

Where you can, paraphrase or remove names and identifying details before using a public tool. Check the privacy settings of any tool you use, and your own organisation's policies.

6. Do you have to tell the funder?

Only some funders ask. The ARC expects applicants to be transparent about their AI use when it requests that information. Other programs may add a question to the form or the declaration. Many don't ask at all.

The rule of thumb is simple: if the guidelines, form or declaration ask about AI, answer honestly. If they don't, you don't need to volunteer it, but you're still accountable for the content. Either way, never use AI in a way you'd be uncomfortable explaining to the funder.

7. A safer way to work with AI

A workflow that keeps you in control:

1. Start with your facts. Write down your project details, figures, dates, partners and evidence before you open any AI tool.
2. Read the guidelines yourself. Note the criteria, weightings, word limits and any AI rules.
3. Draft one criterion at a time, giving the tool only the facts that answer it.
4. Check every claim against your notes and evidence. Delete anything you can't back up.
5. Rewrite in your own voice. Add the local detail and experience only you have.
6. Edit to the limit, then read each answer next to its criterion.
7. Have a person read it before anyone signs the declaration.

For the full application process, see how to write a grant application in Australia, and for answering each question well, how to address the assessment criteria.

8. Before you submit

• You've checked the guidelines and form for any rules about AI.
• Every number, date, name and result is true and you can provide evidence for it.
• No answer contains a claim the AI added that you can't back up.
• Each answer talks about your project specifically, not organisations in general.
• No personal, confidential or commercially sensitive details went into a public AI tool.
• Program details (dates, caps, eligible costs) come from the current guidelines, not from the AI.
• If the funder asks about AI use, you've answered honestly.
• The person signing the declaration has read the final application.

Frequently asked questions

Will my application be rejected if I used AI?

Not because of AI alone, in the published Australian policies we've seen. Funders such as the ARC, NHMRC and City of Melbourne allow AI as a support tool. Applications are still judged on what they say, so vague, generic or inaccurate answers will score poorly whoever wrote them.

Do I have to say I used AI?

Only if your funder asks. Some, like the ARC, expect you to be transparent when they request it, and some forms include a question about it. If the guidelines, form or declaration ask, answer honestly.

Who is responsible if the AI gets something wrong?

You are. Published policies are clear that the applicant is the author and is accountable for the accuracy of everything submitted, including errors that came from an AI tool. Check every fact before you sign the declaration.

Is it safe to put my project details into a public AI tool?

Be cautious. Assume anything you enter could become public. Leave out personal information, confidential figures and unpublished ideas, paraphrase where you can, and check the tool's privacy settings and your organisation's policies first.