Researchers have discovered that some authors are hiding instructions in their manuscripts to manipulate AI-assisted peer review. These 'hidden prompts' can be concealed using white text or microscopic font sizes, and can instruct reviewers to give positive reviews only or follow specific evaluation frameworks. The practice is considered a questionable research practice (QRP) and has been found in 18 academic manuscripts on arXiv. Publishers have inconsistent policies regarding AI use in peer review, with some prohibiting it entirely and others permitting limited use with disclosure requirements. --- Why it matters: This matters to researchers because it highlights the potential for manipulation of AI-assisted peer review processes, which could compromise the integrity of scientific evaluation. It also underscores the need for controlled AI integration in formal review processes and harmonized policies governing AI use in academic evaluation. Source: https://arxiv.org/abs/2507.06185