Beta Version for Detecting AI-Generated Images
Published on March 9, 2025
With the rapid advancements in generative models, detecting AI-generated images in research papers has become more challenging than ever. To assist you in differentiating between real and synthetic content, we have just launched a beta version for detecting AI-generated images in scientific articles.
We developed an AI model to check images in research papers for AI-generated images. Each paper checked by Imagetwin will now also be scanned for AI-generated images. In this post, we focus on the capabilities, limitations, and roadmap for detecting AI-generated images with Imagetwin.
AI-Generated Image Detection in Science
Existing AI detection models are not optimized for scientific images. We tested freely and commercially available APIs and observed low detection rates and high false-positive rates for domain-specific images, such as western blots and microscopic images. To address this, we trained our own detection model using images from research articles.
We generated thousands of AI images using image-to-image, text-to-image, and inpainting (a technique to modify specific parts of an image). We applied transformations like cropping, rotation, and scaling to these images during training to ensure robustness. Examples of AI-generated images used for training and testing are shown below.
We evaluated our model on a test set containing 356 AI-generated images and 292,043 authentic images (published in pre-2022 papers, which we assume are unlikely to be AI-generated). The model correctly identified 271/356 AI-generated images (76% true positive rate) and incorrectly flagged 3,342/292,043 authentic images (1% false-positive rate). When tested on AI-generated images from unseen generative models, the detection rate dropped to 10%, indicating that the model does not generalize well to data outside its training scope.
Examples of AI-generated images detected by Imagetwin
Beta version
Detecting AI-generated images accurately is hard and we want to do it right. Therefore, this feature is an early beta version that can be disabled in your account setting. Below, learn about the limitations and our roadmap to address them.
Coverage of Generative Models
Our detection method is optimized for one of the most accessible and popular model for generating scientific images. However, many AI models are currently available, with new ones emerging monthly. Our results showed that our detection model does not generalize to data outside its training scope, making it less effective at detecting AI-generated images from unseen generative models. In future updates, we focus on expanding training data to include the most common generative models, improving coverage and generalization.
Image Types
The detection model is optimized for domain-specific images commonly targeted for data manipulation, including microscopy images, western blots, histology/pathology slides, cell cultures, and spot images. It is not, however, trained to detect non-scientific photos (e.g., cats, dogs, or unicorns). Future versions will expand training to include a broader spectrum of scientific imagery, such as plots, graphs, and light photography.
Explainability
One major challenge with flagged images is distinguishing false positives from actual AI-generated content. Even a 1% false-positive rate can lead to incorrect detections for papers with many sub-image panels. For future updates, we would like to explore different solutions for increasing explainability. In the meantime, consider the following checks to evaluate detected cases:
- Check for implausible visual cues, such as incomplete or nonsensical text (a common weakness in generative models).
- Review figure labels and textual descriptions for consistency.
- Consider frequency: Multiple flagged images in a paper increase the likelihood of AI-generated content.
- Verify publication dates: Images published before the rise of sophisticated generative models (pre-2022) are unlikely AI-generated.
- Consult authors for original data if uncertainty remains.