TRUST-Preprint: Trustworthy AI-Assisted Peer Review for Preprints

We will deliver an independent, community-respected AI reviewer for preprints that focuses on formal, secondary, and transparency-related aspects rather than replacing human peer review. It generates structured, transparent reports - reporting-guideline checks, methodological reasoning, plausibility tests, literature comparison, and integrity and ethics signals - to help authors and readers judge credibility. Co-designed with scientists-as-citizens, the system operates openly: code, protocols, and datasets are public; evaluation is preregistered; and a controlled pilot with preprint platforms assesses usefulness, reliability, and fairness. Results will reduce the risk of misinformation without delaying the posting and sharing of preprints while supporting faster, more trustworthy open science.