Research funders ‘flooded with AI-assisted applications’

Research agencies risk being overwhelmed by a spike in grant applications from academics using generative artificial intelligence (AI) to craft their funding bids, an analysis suggests.

Drawing on information from 12 research funders across the world, the Research on Research Institute (RoRI) found that the volume of applications in 2025 had risen by an average of 57 per cent compared with 2022, the year in which ChatGPT was launched.

For some prestigious funding programmes, the increase since 2022 is much higher, with applications to the European Union’s Marie Skłodowska-Curie Actions fellowships for early career researchers shooting up by 142 per cent over the three years to 2025.

That was accompanied by a rise in the overall quality of applications, with just 5 per cent of applications falling short of the quality threshold for further consideration in 2025, compared with 20 per cent in 2018, says the study, whose results were drawn from major funders based in Australia, Belgium, Canada, China, Spain, the UK and the European Union, among them the European Research Council, the Australian Research Council and the medical research charity Wellcome.

Writing in Nature, the study’s authors, Geraint Rees, pro vice-chancellor for research at UCL, and James Wilsdon, RoRI’s director, say it is “not yet possible to pinpoint exactly whether or how changes in application volume and quality are linked to AI use”, despite several recent studies indicating that this is strongly likely.

“But if these trends extend to other funders and continue, grant reviewers could soon face huge volumes of high-quality submissions. They will have to make largely arbitrary choices about what or who to fund,” they predict.

To cope with the rise in AI-assisted funding applications, several science funding agencies have introduced bans on the use of GenAI in applications, continue the authors, noting the decision by the US’ National Institutes of Health (NIH) in July 2025 to make applications ineligible for funding if they have been substantially developed by AI.

That approach is, however, “impossible to enforce”, state Wilsdon and Rees, who describe bans as an “inadequate response to the challenge at hand”.

Instead, the authors urge agencies to “consider shifting the emphasis of evaluation away from written proposals, and towards the principal investigator, their research team and their previous and ongoing research programmes”.

Funders should also deploy AI tools to review candidates, using the technology to assess “their publication impact, collaborative reach, methodological diversity and career trajectory”, they add.

“This rich picture would allow funders to compare candidates more completely than can a funding panel. AI could then be used to prioritize and shortlist applications, by comparing data across these profile dimensions to identify candidates whose record is consistent with the statements in their application,” the authors continue.

“Funding panels have always faced hard choices, but they could at least claim to be distinguishing excellent ideas from merely good ones. Agentic AI is making that claim increasingly hollow,” said Rees, the study’s lead author.

“Funders aren’t facing a distant threat – the data suggest the system is already under strain. The good news is that better approaches exist, but the window to act is narrowing.”

The power of new, smarter AI systems is likely to drive application volumes even higher in 2026, commented Wilsdon.

“Meanwhile, peer reviewers will be using the same agentic tools to assess proposals – so we quickly reach a point where systems of grant funding and review will collapse unless funders adopt new strategies for managing volume and demand, and for assessing quality.”

jack.grove@timeshighereducation.com