MDPI signed a multi-year agreement with Proofig AI in September 2025 after a pilot, and began using it first on selected biomedical journals, with plans to extend it to more titles. The editorial team decides whether a manuscript’s image types suit Proofig screening, and the manuscript is uploaded through Proofig’s web interface or an API. Figures are compared with the manuscript itself and with literature indexed in PubMed, producing similarity scores, and a staff member manually reviews every match. MDPI says it tested the available image manipulation detection tools for over a year before choosing Proofig.
Benefits & outcomes
- MDPI and Proofig report that the pilot found duplicated histology and confocal microscopy images, some of them edited to hide the duplication.
- MDPI reports fewer post-publication corrections related to image manipulation.
Notes & quotes
- Proofig’s chief executive Dror Kolodkin-Gal says the pilot achieved “excellent results in detecting problematic images”.
- MDPI prohibits using generative AI to create or enhance research results, including images, graphs and blots.
- “The life sciences are disproportionately affected by research integrity issues.” Tim Tait-Jamieson, Head of Publication Ethics, MDPI.
Publisher / company: MDPIProofig AI
Date of case study: 2025
Status: Live
Built by: Vendor working with a publisher
Sources: