Peer review is a critical research activity, as it has a direct impact on the researchers
applying to DFF. At the same time, peer review in DFF also forms part
of DFF’s decision-making process as a public authority. Furthermore, DFF has an
obligation to protect applicants’ original, unpublished ideas from being exposed to
inclusion in and training of AI models without their consent, thereby protecting
applicants’ intellectual property rights.
For the above reasons, the board has decided that, as a reviewer in DFF, one
must refrain from substantial use of generative AI tools in peer review. It is therefore
not permitted to copy entire applications or parts thereof into, or delegate the
review to, an AI tool.
Ordinary searches for background information do not constitute substantial use
and are permitted in the same way as other literature searches and professional
updating, but must be carried out with due care.
Rephrasing keywords into prose (e.g., in a rejection justification) and using the
tools for linguistic improvement fall into a grey area, as such use is in principle
permitted, but must be carried out with due care to ensure that personally identifiable
information, quotations from the application, or the essence of applicants’
original ideas are not transferred for inclusion in the model.
