AI画像検出器
Upload an image to check for AI-generated content signals across the whole image.
The model analyzes the complete image. It is not a face-liveness, identity, face-swap, deepfake, or tampering detector.
What does AI画像検出器 mean?
AI画像検出器 checks an uploaded image for signals associated with AI-generated content. The complete image is analyzed, including subjects and background.
What ImgAuth detects
- Uploads the original image whenever it fits the provider limit
- Analyzes AI-generation signals across the complete image
- Returns a policy result and model risk score
What ImgAuth does not detect
- Editing, resizing, screenshots, and recompression can affect the score
- This is not a face-liveness, identity, face-swap, deepfake, or tampering detector
- Not a legal identity verification, KYC, ownership, or provenance workflow
Inputs
- portrait image
- scenery or product image
- artwork or illustration
- other supported image upload
Outputs
- 判定結果
- AI生成スコア
- 解析対象画像
How AI画像検出器 works
- Uploads the original image whenever it fits the provider limit
- Analyzes AI-generation signals across the complete image
- Returns a policy result and model risk score
- Supports portraits, scenery, products, artwork, and other image subjects
Use Cases
- Screening uploaded images for AI-generated content
- Reviewing portraits and profile images for whole-image AI signals
- Checking scenery, product, artwork, or social images
- Supporting moderation and trust-review workflows
Important capability boundary
The model analyzes the complete image. It is not a face-liveness, identity, face-swap, deepfake, or tampering detector.
Limitations
- The result is a model signal, not proof of origin or authenticity
- Editing, resizing, screenshots, and recompression can affect the score
- This is not a face-liveness, identity, face-swap, deepfake, or tampering detector
- Borderline results should be reviewed with the image context and other evidence
FAQ
What does AI画像検出器 detect?
It detects model signals associated with AI-generated content across the complete uploaded image.
Does it only analyze faces?
No. It analyzes the complete image, including people, objects, scenery, artwork, and backgrounds.
Can it prove that an image is real?
No. A low score means the model found weaker AI-generation signals; it does not prove authenticity or origin.
Can it detect face swaps or image tampering?
Not specifically. This workflow is for whole-image AI-generation signals, not face liveness, identity, face-swap, deepfake, or tampering analysis.
What does the AI Generation Score mean?
AI Generation Score is Tencent's model risk score for AI-generated content across the whole image. It is not a calibrated probability and does not prove that an image is authentic.