On January 17, 2025, Tencent launched an AI-generated image detection tool aimed at helping people identify whether an image was created by AI.
The AI-generated image detection system developed by Tencent's Phoenix Lab in the Mixed Yuan Security Team can determine if an image is AI-generated within seconds by uploading the image and waiting for verification. The system primarily distinguishes between real images and AI-generated ones by capturing differences, such as AI images sometimes lacking common sense logic, resulting in unrealistic content like a flying puppy with wings or a cat smoking a cigar; AI-generated images often require "watermarking," meaning they must include explicit or implicit markers, which may be visible to the naked eye or not, requiring detection tools to read; AI-generated images also contain some invisible hidden features, such as bright spots with densely distributed local textures when adjusting the HSV color space.
Identifying AI-generated images cannot rely solely on a single criterion; the detection system must utilize AI models to capture various feature differences, including image texture, semantics, and hidden characteristics. The Phoenix Lab used 1.4 million positive and negative samples for model training, covering a variety of generated content scenarios, achieving a final detection rate of over 95%, with ongoing optimization and improvement. The Phoenix Lab also developed a text detection system that learns from massive data to achieve text detection, collecting a large number of positive and negative samples for training, and using comparative methods to infer the probability of AI generation in articles, enhancing detection capabilities for unseen data. Currently, this system encompasses diverse styles such as news reports, official documents, novels, and essays, with plans to strengthen detection for poetry and other genres in the future.
The initial purpose of developing AI detection tools at the Phoenix Lab was to optimize the training corpus for the Mixed Yuan large model. However, with the explosive growth of AI, the demand in academia and industry for "using AI to detect AI-generated content" has increased, leading the lab to develop two tools: AI-generated image detection and AI-generated text detection, providing experiential services.
In the future, a detection tool for AI-generated videos will also be launched to further enhance detection capabilities.
Experience Address:
https://matrix.tencent.com/ai-detect/ai_gen
https://matrix.tencent.com/ai-detect