New Feature To Tackle Fake Images in Science

Further strengthening its commitment to preserving image integrity in scientific research, the founders behind image integrity tool Proofig AI have released the AI Image Fabrication identifier tool, an AI-powered detection feature that specifically targets the detection of AI-generated images. By detecting these images, the tool will alert users to investigate. This will help ensure the authenticity and integrity of published images in scientific research — addressing an imminent and prominent potential problem for science.

The use of AI-generated images has risen dramatically due to advancements in the technology and accessibility of image models and is used to generate images quickly and efficiently. It has been recommended that Generative AI can be used as a basic tool to support authors in refining, formatting and correcting texts — but must not be used create, alter or manipulate research data and results, including images[1].

The new feature will alert users to microscopy images that might be AI-generated and warrant further investigation when scanning manuscripts. Proofig AI combines machine learning, pattern recognition and statistical analysis to detect anomalies in images that suggest AI generation. The system is trained on a vast dataset of known AI-generated images to recognise subtle differences that may not be apparent to the human eye. Users can take the findings of the report and images detected for potential signs of manipulations, investigating further to determine how to proceed.

“The detection of AI-generated images will become increasingly crucial as AI technology continues to advance,” said Dr. Dror Kolodkin-Gal, Founder of Proofig AI. “The aim of the feature is to further aid researchers and publishers with an additional layer of scrutiny, helping to prevent the publication of manipulated or fraudulent images.

“Alerting users (editors and researchers) to investigate images before any scientific publication and can help protect the integrity of individual studies and the reputation of respected academic institutions and publishers, whilst enhancing the trust of scientific literature.”

Proofig AI’s new detection feature is the first tool to detect instances of AI-generated microscopy images within the scientific research community. The new feature integrates into the existing Proofig AI platform and uses advanced algorithms to adapt to new AI models as they emerge, allowing users to stay ahead of potential image manipulation threats while providing a smooth and automated process to verify image authenticity as part of their regular workflow.

The release of the AI Image Fabrication feature follows the recent release of Proofig AI’s latest plagiarism tool update. The tool now allows users to compare manuscripts against existing papers with a database of tens of millions of images. This feature has increased image capacity, enabling users to generate reports with up to 500 subimages for in-depth analysis, along with smarter algorithms to the reduce the rate of false positives.