Using AI for Grant Writing: A Practical Guide for Australian Nonprofits

If you've typed a grant application into ChatGPT and felt vaguely guilty about it, you're not alone. Almost every sustainability-focused nonprofit and social enterprise I speak with is doing some version of this.

Australian funders are not penalising the use of AI in grant applications, but they are rejecting applications where AI has replaced the applicant's genuine program knowledge, community insight, and mission voice. And any organisation using an AI chat, ChatGPT, Gemini, or Claude, to process grant application content that includes beneficiary data must consider APP 8 cross-border disclosure obligations under the Privacy Act 1988, regardless of the application's funding outcome.

That's two separate issues. Most of the advice out there only addresses the first one.

Will Australian funders penalise you for using AI in grant applications?

Based on current evidence: not directly.

Research published by Equitable Philanthropy in October 2025, presented at the Charities and Not-for-Profits Conference, found that most Australian funders are comfortable with AI tools including ChatGPT and sector-specific tools like Drafter from Funding Centre. Very few currently require AI disclosure in applications.

The caveat is that funders are increasingly able to recognise generic AI-generated language, and they are rejecting those applications. Not because AI was used, but because it was used badly.

As Amy Waters from the Geelong Community Foundation put it: "It is essential that charities do the initial work to clearly define the activity that they are seeking funding for. This cannot be AI driven. It needs to be informed by the applicant's understanding of their target group and their real-world experiences."

The decision when writing a grant should not be to use AI versus not use AI. It's whether the application demonstrates genuine program knowledge, community insight, and a clear theory of change, or whether it reads like something assembled from a prompt and a template.

The Australian Research Council has also introduced specific controls on generative AI use in grant programs. That signals where higher-stakes funding environments may be heading.

What do Australian funders actually think about AI-generated grant applications?

The Equitable Philanthropy research is the most useful data point currently available for Australian nonprofits. The headline finding is reassuring: funders aren't conducting AI witch-hunts.

But the nuance matters. Gill Whelan from the Decjuba Foundation described the right approach well: "AI tools should be used as a starting point only. Make sure you imbue your writing with the unique tone of your organisation to avoid sounding like everyone else."

What funders are looking for, and what AI cannot generate, is lived experience, mission voice, and authentic program logic grounded in your specific community context. An AI can help you structure an argument, improve a sentence, or draft a first pass. It cannot tell the funder what you know about the people you serve.

The research from SmartyGrants and Grant Assistant echoes this: the problem isn't AI use, it's AI replacement. When the tool writes the whole thing and the organisation just signs off, the application loses the substance that funders are actually evaluating.

What's the difference between using AI well and getting your application rejected?

The organisations using AI effectively in grant writing are doing four things that distinguish them from those getting rejected.

First, they do the program work before touching the AI. The theory of change, the program logic, the community data, and the outcome evidence come from the organisation, not from prompting. The AI receives specific, detailed input, and that input quality determines the output quality.

Second, they use AI for structure and language, not for substance. A well-designed prompt can turn a solid set of notes into a first draft that reads clearly and covers the required sections. That is genuinely valuable. The substance, why this program, why this community, and why now, still comes from the team.

Third, they review for mission voice. The final application sounds like the organisation, not like a general writing tool. If someone removed your name from the application and a funder could not tell who wrote it, the AI has done too much work.

Finally, they're deliberate about which tools they use for which tasks. More on that in the data risk section below.

How do you maintain your organisation's voice when using AI for grant writing?

The organisations that maintain their authentic voice in AI-assisted grant writing have built an internal library before they sit down to write: previous applications that worked, descriptions of programs written by frontline staff, quotes from beneficiaries, and language that's specific to their organisation's way of describing their work.

This source material becomes the AI's context. You feed it your language, your frameworks, your evidence, and you ask it to help you structure an argument, not to generate one from scratch. The difference in output quality is significant.

It's also worth naming that grant writing is one of the highest-stakes use cases for an NGO's voice. A generic-sounding application does not just risk rejection. It signals to funders that your organisation may not have the program depth or community connection the funding is designed to support.

If AI is making your applications sound less like you, that's feedback about your process, not about AI tools in general.

What data risks come with using AI tools for grant applications?

Grant applications often contain sensitive information: program outcome data, descriptions of beneficiary situations, demographic data, and case studies drawn from real people's experiences. When that content is entered into a free plan of an AI chat such as ChatGPT or Gemini, it may be processed or stored outside Australia depending on the tool and account settings.

Under APP 8 of the Privacy Act 1988, organisations that disclose personal information to overseas recipients are generally required to take reasonable steps to ensure the recipient handles it in accordance with the Australian Privacy Principles, unless an exception applies.

This isn't a theoretical risk. It's a compliance question that your organisation should be able to answer.

At minimum, your team should know which AI tools are approved for which types of content, and what categories of information should not be entered into any AI tool without additional controls.

How do you build an AI-assisted grant writing process that reflects your organisation's genuine expertise?

Here's what most grant writing AI advice misses: the quality of your AI-assisted grant applications is a direct reflection of whether your organisation has thought through its AI use systematically.

A social enterprise that has an AI acceptable use policy, has decided which tools are safe for which data, and has trained its team on responsible AI use will write better grant applications using AI, because the team knows what their organisation can and cannot credibly claim. The governance work and the quality work are the same work.

The organisations that struggle with AI-assisted grant writing tend to have the same problem: different people are using different tools in different ways, with no shared framework for what AI should and should not do. One person uses ChatGPT for the whole application; another will not touch it; a third is manually checking whether anything looks too polished. The result is inconsistency, risk exposure, and applications that do not reflect the organisation's actual knowledge.

An AI governance policy does not need to be long or complicated. For a small organisation, it can cover the essentials in two pages: which tools are approved, which data categories are off-limits for AI processing, what human review looks like before anything goes out the door, and who's responsible for keeping the policy current as tools change.

When your team has that clarity, using AI for grant writing becomes a straightforward productivity gain rather than an ongoing source of anxiety. The application reads like your organisation because your team knows how to use the tools in a way that supports rather than replaces their expertise.