StealCut AI Image Detector Achieves 95.2% Accuracy

StealCut, an AI security startup, has launched a new AI image detection model. This model achieves 95.2% accuracy. It has outperformed three leading global detection services in benchmark testing conducted in March 2026. The benchmark tested AI-generated images from thirteen new generative AI models, including Grok, Gemini, and ChatGPT. StealCut surpassed its competitors in overall accuracy, forged-image detection rate, and average detection performance across these models. What's interesting is that in a June 2026 benchmark, the model recorded a 91.7% detection rate for forged images. It also achieved a 96.5% average detection rate across the latest generative AI models. StealCut’s technology is also being used in a project with SBS, a major broadcaster in South Korea. The company aims to address the challenges of increasingly realistic AI-generated images that are difficult for humans to identify. This technology matters because it helps newsrooms, online platforms, and businesses flag manipulated images before they can cause harm like misreporting, fraud, or reputational damage.