TRU.th
TRU.th is ATNA's standalone deepfake and media authenticity detection API. Submit any image or video and TRU.th determines whether it is AI-generated, synthetically altered, or genuine. Built for high-volume media verification pipelines.
Why TRU.th?
- Detect AI-generated faces — identifies GAN, diffusion model, and face-swap outputs.
- Video deepfake detection — frame-by-frame analysis to catch synthetic video content.
- Image manipulation detection — surfaces inpainting, splicing, and generative fill edits.
- Model-agnostic — detects Midjourney, DALL-E, Stable Diffusion, and other major generators.
- High-volume API — designed for media moderation and KYC screening at scale.
Capabilities
1. Deepfake Probability Score
0–100 confidence that the submitted media is AI-generated or manipulated.
2. Manipulation Type Classification
Identifies the manipulation type: face swap, GAN-generated, diffusion-generated, or inpainted region.
3. Video Frame Analysis
Per-frame scoring with aggregate video verdict. Supports MP4, MOV, and AVI.
4. Metadata Forensics
Detects stripped or inconsistent EXIF data commonly associated with synthetic media.
How It Works
- Upload image or video to the TRU.th endpoint.
- AI models analyze for synthetic artifacts, manipulation traces, and metadata anomalies.
- Returns deepfake probability, manipulation type, and confidence level.
Use Cases
1. KYC selfie deepfake screening Detect AI-generated selfies submitted during remote onboarding before human review.
2. Social media and UGC moderation Screen user-uploaded content at scale to detect synthetic media violating platform policies.
3. Insurance claim photo fraud Detect manipulated vehicle, medical, or property photos in claim submissions.
4. Legal evidence verification Verify authenticity of digital evidence — screenshots, photos, and videos — before legal proceedings.
View API Reference →