Dutch Research Council Updates Generative AI Policy for Research Grant Applications and Evaluation Process

Dutch Research Council Updates Generative AI Policy for Research Grant Applications and Evaluation Process

(IN BRIEF) The Dutch Research Council (NWO) has updated its policy governing the use of generative artificial intelligence in its grant application process. Under the revised guidelines, applicants may use generative AI tools when preparing proposals, but assessors and NWO staff remain prohibited from using such tools during the evaluation process due to confidentiality requirements and verification responsibilities. The updated policy introduces several clarifications, including a ban on hidden AI prompts in applications, simplified disclosure requirements for applicants who use AI as a source, and formal declarations from referees confirming they have not used AI. The policy also acknowledges the current limitations of AI detection software and provides guidance on sustainability considerations when deciding whether to use AI tools. While AI remains restricted in grant evaluations, NWO staff may use approved applications such as Microsoft Copilot Chat and DeepL for other work activities.

(PRESS RELEASE) THE HAGUE, 17-Mar-2026 — /EuropaWire/ — Dutch Research Council has updated its policy on the use of generative artificial intelligence (GAI), providing clearer guidance for applicants, assessors and staff involved in the grant application process. The revised policy confirms that applicants are permitted to use generative AI tools in preparing their proposals, while the use of such technologies remains prohibited for assessors and NWO employees during the evaluation of funding applications.

The council first introduced guidelines for generative AI use last year. Given the rapid pace of developments in artificial intelligence and the practical experience gained since the initial rollout, NWO has continued refining its framework to ensure that the rules remain relevant and effective.

Under the updated policy, the use of generative AI is still completely excluded from the assessment phase of grant applications. Referees, committee members, jury members and NWO staff involved in evaluating proposals must follow strict confidentiality requirements for all application materials. According to NWO, these confidentiality obligations cannot currently be guaranteed if generative AI tools are used in the evaluation process. The council also has a responsibility to verify that assessments are based on careful, independent analysis—something that may be difficult to confirm when AI-generated assistance is involved.

Several updates have been introduced to the policy framework. One addition addresses the limitations of AI detection tools. NWO notes that current software designed to identify AI-generated content is not sufficiently reliable, meaning results from such tools will not be accepted as evidence of AI use in grant applications.

The updated policy also clarifies how sustainability considerations apply to the use of AI. NWO encourages staff and participants in the grant process to consider whether AI tools are genuinely necessary or whether more sustainable alternatives could be used.

For applicants, the guidance now explicitly prohibits the inclusion of hidden prompts in grant proposals. Hidden prompts are instructions embedded in documents that are difficult to detect by human readers but are designed to influence AI systems—for example by prompting a tool to produce favourable assessments of the text. Such practices are not allowed under any circumstances.

Transparency requirements have also been simplified. Applicants are only required to disclose generative AI when it is used as a source within their application, in which case it should be cited in the references section.

The policy further clarifies rules for referees and assessors. The ISAAC referee form now requires reviewers to formally confirm that they have not used generative AI during their evaluation. The guidelines also state explicitly that reviewers may not use AI tools even for language editing or translation assistance.

NWO employees are similarly prohibited from using generative AI tools during the grant assessment process. The updated guidance also explains how staff should interact with external partners who may wish to use AI tools, emphasizing the need to prevent confidential information from being shared with such systems.

Outside of grant evaluation activities, NWO staff may use approved generative AI applications for other professional tasks. However, only tools that have been formally vetted through the organization’s IT procurement process are permitted. Currently approved applications include Microsoft Copilot Chat and the DeepL translation service.

Media Contact:

Freek Hofland
Spokesperson
+31 6 58920086
f.hofland@nwo.nl

SOURCE: NWO