UK Research Funders Revised Their Joint Generative AI Statement in June 2026

In September 2024 UK Research and Innovation published its policy on the use of generative artificial intelligence in application preparation and assessment. For applicants, it set an expectation of transparency: in line with sector good practice, applicants were expected to be transparent where they had used generative AI tools in developing an application, and UKRI said that such disclosure would not affect assessment.

For assessors, the policy was strict. Reviewers were prohibited from using generative AI tools to develop their reviews of funding proposals, and a secondary summary reports that the prohibition extended to any part of assessment, including language or grammar correction. UKRI treated the release of confidential material into generative AI tools as a breach of confidentiality and integrity, and stated that it could reject applications or bar individuals from future submissions where misconduct is upheld.

The UKRI policy reflected a 2023 joint statement by UK funders that had barred AI from peer review outright.

In June 2026 a revised joint statement was published on behalf of the Association of Medical Research Charities, the British Heart Foundation, Cancer Research UK, the Royal Academy of Engineering, the Royal Society, UK Research and Innovation and Wellcome. The main change is that it now addresses funders’ own use of generative AI. Speaking before publication, a UKRI representative said funders “may use generative AI to process funding applications”, but would not put a whole application into a publicly available tool. The statement says that final funding decisions are underpinned by human oversight and that funders do not use generative AI tools to make funding decisions.

For applicants, the statement sets two limits. Applicants must not use generative AI tools for uses that would breach their standard research integrity expectations, or to generate an entire application, or sections of one, without human involvement. Acknowledging generative AI outputs is described as best practice.

For peer reviewers, the statement says that they must not input confidential application or review content into generative AI tools. This is narrower than the 2023 position, which barred AI from peer review altogether. Pubrica did not obtain the complete text of the revised statement, and reports differed on the exact timing of publication, so readers should check the funders’ published statement for the precise wording.

The joint statement sets a common baseline, but funders apply their own detailed rules. Wellcome, for example, requires applicants to declare the use of generative AI tools when applying for grant funding, with some exceptions set out in its policy, which was last updated in April 2026. UKRI’s September 2024 policy remains published on its website, and applicants and reviewers should check whether UKRI has updated it to reflect the 2026 joint statement before relying on its stricter reviewer provisions or the more permissive joint wording.

Universities have issued their own guidance to staff preparing bids. Research support offices at institutions such as Cambridge and Liverpool provide advice on how researchers may use AI in grant writing within funder rules.

The UK approach is broadly in line with recent changes in Australia, where the ARC and NHMRC have since 28 April 2026 allowed reviewers to use AI to refine the wording of comments while requiring their own judgement for scoring. It differs from the position of the US National Institutes of Health, which continues to prohibit its peer reviewers from using generative AI to analyse applications or formulate critiques. Researchers who apply to or review for funders in several countries should check each funder’s rules separately.

Applicants should ensure that the ideas, design and substance of a proposal are their own, use AI only with active human involvement, and acknowledge AI use where the funder requires or recommends it. Reviewers should never paste application text or their draft reviews into public generative AI tools, and should follow each funder’s specific rules on any permitted assistance.

Grant writing consultants and medical writers supporting UK applications should agree with applicants how AI will be used, record that use, and check the declaration requirements of each funder. This brief summarises published funder positions for information and is not affiliated with or endorsed by any funder.