Proofig

Overview

Proofig is an AI image integrity detection platform designed for scientific publishing. Using advanced computer vision and deep learning, it automatically identifies image manipulation, duplication, plagiarism, and AI-generated images in academic manuscripts. It is widely adopted by leading research institutions, publishers, and journals, including Science journals and the Nature Publishing Group, making it a standard tool for ensuring the authenticity of scientific images.

The platform supports multiple detection types: traces of manipulation within a single image (such as cloning, splicing, deletion), duplication of images/charts/figures within a manuscript, plagiarism comparison against a database of over 155 million publicly available images from PubMed, and detection of images generated by the latest AI models. Proofig also offers a "My Database" feature allowing researchers to build their own historical image library to prevent self-plagiarism. Detection accuracy reaches 99.4% with an extremely low false positive rate.

Proofig's workflow is exceptionally simple: after uploading a manuscript, all images are automatically analyzed, suspicious areas are flagged, and forensic-level tools are provided for manual review, finally generating a detailed report. The entire process takes only a few minutes, greatly improving the efficiency of editors and authors while avoiding retractions and reputation damage caused by image issues. It has been validated on millions of sub-images and hundreds of thousands of manuscripts, and is trusted by Nobel laureates and research integrity experts.

Core capabilities

Image tampering detection

Identifies traces of cloning, editing, deletion, splicing, and other manipulations

Duplicate use analysis

Detects duplication and reuse of images, charts, and figures within a manuscript

AI image recognition

Accurately identifies fake images generated by mainstream AI models

Image plagiarism comparison

Compares similarity against a database of 155 million+ PubMed images

Self-plagiarism prevention

Prevents authors from reusing their own published images via a personal historical database

High accuracy, low false positives

Image extraction accuracy 99.4%, OCR accuracy 96.8%, and AI image detection false positive rate below 0.01%

Fully automated workflow

Upload and analyze, results in minutes

Forensic-grade review tools

Provides zoom, comparison, and annotation tools to assist manual judgment

Confidentiality assurance

Zero data reuse, ensuring manuscript privacy and security

Use cases

Researchers use Proofig to self-check before submission, avoiding unintentional errors that lead to retraction

Journal editors conduct routine image integrity screening on submitted manuscripts

Publishers batch-detect large volumes of manuscripts to ensure publication quality

Research institutions pre-review internal papers before publication to protect institutional reputation

Ethics committees obtain forensic-grade evidence when investigating image misconduct

Authors use the 'My Database' feature to prevent self-plagiarism, ensuring each paper's images are used for the first time

Reviewers leverage automated tools to quickly flag suspicious images, improving review efficiency

University labs incorporate Proofig into research integrity training to raise students' awareness of standards