#Peerreview

A new controversy has erupted in academic circles after investigators uncovered that a group of international researchers embedded secret instructions—so-called “hidden AI prompts”—within preprint manuscripts to influence AI-powered peer review systems toward more favorable feedback. The revelations were detailed in recent reports, following a data-driven exposé that found 17 preprint articles on the arXiv platform with covert commands instructing AI models to deliver only positive reviews, avoid criticism, and even explicitly recommend the work for its novelty and methodological rigor. This manipulation was achieved through invisible white text or minuscule fonts, remaining undetected by human readers but fully readable by AI engines tasked with the review process (Nikkei Asia, ExtremeTech, Japan Times).

A global scare highlights how covert AI prompts in preprint papers can steer reviews toward praise. Investigations found 17 arXiv preprints containing hidden instructions that urged AI review models to be positive, avoid criticism, and even support publication for novelty and rigor. The prompts used barely legible text unread by humans but readable by AI.

This issue matters beyond a technical trick. As large language models become more common in research tasks, the integrity of peer evaluation could be at risk. For Thailand, where academic quality underpins national progress and international collaboration, the incident raises questions about safeguarding trust in science amid sophisticated digital manipulation.