arXiv:2606.10159v1 Announce Type: new Abstract: AI is increasingly used to support scientific peer review, from manuscript screening, reviewer assistance to editorial triage. Although such systems promise to reduce reviewer burden and accelerate publication, their robustness to strategic manipulation remains poorly understood. Here we show that AI-mediated peer review is vulnerable to a simple, low-cost manipulation: superficial rephrasing of the manuscript abstract. Without changing the underlying scientific content and communication, and even without knowledge of the reviewing model, adversa
The increasing integration of AI into scientific peer review systems makes its vulnerabilities a pressing concern.
This highlights a significant risk to the integrity and credibility of scientific publishing, potentially undermining trust in research.
The perceived infallibility of AI in critical academic processes is challenged, necessitating new security protocols for AI-assisted systems.
- · Cybersecurity researchers
- · AI robustness solution providers
- · Manual peer review advocates
- · AI-assisted peer review system developers
- · Academic publishers relying solely on AI for screening
- · Researchers using manipulative tactics
Increased scrutiny and reevaluation of AI's role in sensitive academic processes will occur.
Development of more sophisticated, adversarial-resistant AI peer review tools will accelerate.
A potential chilling effect on AI adoption in other high-stakes decision-making environments could emerge, due to these identified vulnerabilities.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL