Should Applicants Be Told if AI Was Used to Review Their Grant Proposal?

Do Researchers Have a Right to Know Whether AI Evaluated Their Grant Application?

You submit a research grant proposal and receive an unfavorable evaluation. The reviewers question your methodology, describe your contribution as insufficiently original, and assign a score below the funding threshold.

Later, you discover that an AI system may have helped generate some of those assessments.

Would knowing this information have changed how you interpreted the review? Should the funding organization have disclosed the AI involvement before the evaluation, after the decision, or at all?

These questions become particularly important when AI moves beyond administrative assistance and contributes to scientific judgments, numerical scores, applicant rankings, or recommendations about which projects deserve funding.

The issue is not simply whether AI was used. It is whether applicants receive enough information to understand how their proposals were evaluated and who remains accountable for the resulting decisions.

Meaningful AI Involvement Should Generally Be Disclosed

What Transparency Should Mean in AI-Assisted Grant Review

Why AI Disclosure Matters to Grant Applicants

Research grant evaluation is a consequential process. Funding decisions can determine whether a project proceeds, whether researchers can recruit staff, and whether early-career investigators can establish independent research programs.

Applicants generally understand that grant proposals may undergo administrative screening, external peer review, panel discussion, and funding decisions based on published criteria. The precise procedures vary among funding organizations.

When AI becomes part of that process, applicants may reasonably want to know whether their work was evaluated by human experts, an automated system, or some combination of the two.

Disclosure helps distinguish the source of an assessment from the authority responsible for it. It can also make it easier to identify which parts of a decision may require additional scrutiny.

However, transparency should not be confused with a guarantee of fairness. An organization can openly disclose an AI system while still using it in ways that are scientifically inappropriate or inconsistent with its own policies.

Not Every Use of AI Has the Same Disclosure Significance

The case for disclosure becomes stronger as AI moves closer to substantive evaluation and consequential decision-making.

For example, an automated system that checks whether required application fields are complete performs a different function from a language model that evaluates scientific originality.

These distinctions describe the potential significance of different AI functions. They are not universal classifications established by every funding agency.

Even seemingly administrative systems may warrant additional disclosure if their errors can prevent an otherwise eligible proposal from receiving scientific review.

Some Research Funders Prohibit AI Use in Scientific Review

Before asking how AI involvement should be disclosed, researchers should establish whether the funding organization permits that involvement.

The U.S. National Institutes of Health (NIH) issued Notice NOT-OD-23-149 in June 2023, prohibiting scientific peer reviewers from using generative AI technologies to analyze and formulate critiques of grant applications and research and development contract proposals.

The notice also explains that uploading or sharing confidential application content or original concepts with online generative AI tools violates NIH peer-review confidentiality requirements. The official NIH notice provides the governing requirements for the activities it covers.

The European Research Council (ERC) clarified its guidance in March 2026. Its principles include non-delegation of evaluation and protection of confidentiality. ERC reviewers may not use AI to summarize proposals in place of reading them, assess scientific merit, or generate draft evaluations. Limited uses, such as improving review language without sharing proposal content, are permitted under specified conditions.

These examples demonstrate that research funders may draw different boundaries around acceptable AI assistance. They do not establish a universal policy for all national funding agencies, private foundations, or institutional research programs.

Where a particular AI use is prohibited, informing applicants that it occurred does not resolve the underlying violation.

What Information Should Applicants Receive About AI Involvement?

A meaningful disclosure should explain the role of AI in terms that applicants can understand. Simply stating that an organization uses advanced technologies may provide little useful information about the actual evaluation process.

Where AI materially contributes to assessment and its use is permitted, a disclosure could address the following questions:

• What stage of the review process involved AI?
• What task did the AI system perform?
• Did it generate summaries, criticisms, scores, rankings, or recommendations?
• How were its outputs checked before influencing the evaluation?
• Who retained responsibility for the final assessment?
• What procedure is available for raising concerns about a possible error?

The level of technical detail may depend on the system's role, applicable law, security considerations, and the need to protect confidential evaluation materials.

Applicants do not necessarily need access to model weights, private reviewer deliberations, or confidential information about competing proposals to receive a meaningful explanation of AI involvement.

Should AI Use Be Disclosed Before or After the Funding Decision?

Disclosure timing affects its usefulness.

Advance notice allows applicants to understand the evaluation procedure before submitting their work. It may also inform their decisions about participation, particularly when proposals contain unpublished intellectual property or commercially sensitive information.

Post-decision disclosure can help applicants interpret reviewer feedback and identify possible grounds for requesting clarification or correction.

For consequential AI involvement, a combination of general advance notice and decision-specific information may be more informative than either approach alone.

A general statement that AI may be used does not necessarily tell an applicant whether their own proposal was assessed with AI assistance.

Conversely, application-specific disclosure may require reliable internal records showing when the system was used and what role its outputs played.

Does Disclosure Require Revealing the Identity of Human Reviewers?

Not necessarily. Transparency about AI involvement and confidentiality of peer-review deliberations address different concerns.

Many funding systems protect reviewer identities or restrict access to internal deliberations to support candid evaluation and manage conflicts of interest.

An organization may be able to explain that AI assisted with an authorized administrative task without identifying individual reviewers or revealing confidential discussions.

Similarly, an applicant may receive information about the existence and function of an automated scoring system without gaining access to another applicant's proposal.

Disclosure arrangements should preserve legitimate confidentiality protections while providing sufficient information to understand the evaluation process.

Can AI Disclosure Help Applicants Challenge an Unfair Evaluation?

Disclosure may make it easier to identify possible errors, but it does not automatically create a legal right to appeal a funding decision.

Appeal and reconsideration procedures differ among funding organizations. Some permit challenges involving procedural errors or conflicts of interest but do not reconsider scientific judgments merely because an applicant disagrees with them.

Where an AI system materially contributed to an assessment, applicants may reasonably ask whether its outputs were verified and whether the procedure complied with the funder's published rules.

For example, if an AI-generated criticism incorrectly claims that a proposal lacks a methodology section, the issue may concern an identifiable factual or procedural error rather than an ordinary disagreement about scientific merit.

Any request for correction should follow the funding organization's established communication and review procedures.

Why Disclosure Alone Cannot Prevent AI Bias

An AI system may systematically favor particular institutions, research approaches, or styles of academic writing. Disclosing its use does not eliminate those tendencies.

Likewise, a reviewer may become influenced by an AI-generated score even when the organization formally assigns final responsibility to human evaluators.

These concerns connect to bias in AI-assisted grant scoring and the possibility that automated systems may reproduce historical inequalities in research funding.

Meaningful governance therefore requires more than a disclosure statement. It also requires appropriate validation, monitoring, and procedures for identifying and correcting consequential errors.

The NIST AI Risk Management Framework provides a general approach to AI governance, including transparency, explainability, accountability, and the management of harmful bias. Its principles can inform research funding practices, although they do not establish one mandatory disclosure format for all grant programs.

Who Remains Responsible When AI Contributes to a Funding Decision?

An AI system cannot assume institutional responsibility for allocating research funds.

The funding organization remains responsible for establishing appropriate evaluation procedures, applying its published criteria, and complying with applicable requirements.

Human reviewers and decision-makers also retain responsibilities determined by their roles and the organization's policies.

This remains true even when an AI system produces the initial summary, identifies possible weaknesses, or generates a numerical recommendation.

The relevant governance question is which decisions may appropriately receive AI assistance and which require independent human judgment.

When a Grant Applicant Discovers That AI Contributed to Their Evaluation

The scenario is hypothetical and does not describe an actual foundation or documented funding dispute.

It illustrates why meaningful disclosure should be accompanied by traceable evaluation records and procedures for examining potential errors. Without those safeguards, applicants may know that AI was involved but have little practical means of determining whether its involvement affected their assessment.

Misconceptions About AI Disclosure in Grant Review

What Applicants and Funding Organizations Should Do About AI Disclosure

For applicants, the most useful starting point is the funding organization's official evaluation policy.

Before submission, examine whether the organization explains its use of automated systems, confidentiality protections, and review procedures. If the policy is unclear and AI involvement would materially affect your decision to submit, seek clarification through an authorized channel.

After receiving a decision, distinguish between disagreement with a reviewer's scientific judgment and evidence of a possible factual or procedural error.

For funding organizations, disclosure should be proportionate to the influence of AI on the assessment and sufficiently specific to support accountability.

Disclosure Should Be Supported by an Internal Audit Trail

A funding organization cannot reliably explain AI involvement if it has not documented how the system was used.

Where AI assistance is permitted, relevant records may include the system's function, version, authorized inputs, generated outputs, validation results, and any human decisions that relied on those outputs.

These records should be retained according to applicable governance, confidentiality, and records-management requirements.

They may also support investigations into confidentiality risks associated with AI-assisted grant review.

The objective of disclosure is not to make AI use inherently acceptable. It is to ensure that applicants understand the relevant evaluation process and that funding organizations remain answerable for the decisions they make.

Checklist for Transparent AI Use in Grant Evaluation

Frequently Asked Questions About Disclosing AI Use in Grant Review

Applicants Should Understand When AI Influences Their Evaluation

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.