TruthScan API · v2.1

TruthScan API
for Developers

Detect AI-generated text, images, audio, and video in your own product. REST API with JSON responses, webhooks, and enterprise SLA.

Request API KeyView Endpoints ↓

2.8 s

Avg response time

99.9%

Uptime SLA

11

Models in ensemble

50+

Languages supported

Authentication

All API requests require a Bearer token in the Authorization header.Request an API key →

Authorization: Bearer ts_live_xxxxxxxxxxxxxxxxxxxxxxxx

Endpoints

POST/api/detect/text

Detect AI-generated text. Supports 50+ languages.

Request body: { "text": "string" }

Example

curl -X POST https://aigeneratedit.com/api/detect/text \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{"text": "The quick brown fox jumps..."}'
POST/api/detect/file

Detect AI-generated images, audio, or video. Multipart upload.

Request body: FormData: file (blob), type ("image" | "audio" | "video")

Example

curl -X POST https://aigeneratedit.com/api/detect/file \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "file=@image.jpg" \
  -F "type=image"
POST/api/detect/url

Multi-modal scan of any public URL — text, images, audio, and video.

Request body: { "url": "string" }

Example

curl -X POST https://aigeneratedit.com/api/detect/url \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{"url": "https://example.com/article"}'

Response Format

All endpoints return the same standardized JSON structure:

{
  "verdict": "ai",
  "ai_probability": 0.91,
  "human_probability": 0.09,
  "confidence": 94,
  "signals": [
    { "label": "Low perplexity score — GPT-family smoothness", "severity": "high" },
    { "label": "Burstiness coefficient 0.09 — below human baseline", "severity": "high" },
    { "label": "GPT-4o model signature detected (71% confidence)", "severity": "medium" }
  ],
  "model_used": "TruthScan Text v2.1 — RoBERTa + Binoculars + Fast-Detect-GPT",
  "processing_time_ms": 1243
}
verdict

"ai" | "human" | "mixed"

Final classification

ai_probability

number (0–1)

Probability this is AI

confidence

integer (0–100)

Model certainty %

signals

Signal[]

Forensic evidence list

model_used

string

Models that ran

processing_time_ms

integer

Latency in milliseconds

TruthScan AI Models

All models run in ensemble. Results are voted and weighted by confidence.

ModelTypeAccuracyDescription
RoBERTa-base-OpenAI-detectorText93%Hugging Face fine-tuned classifier
Binoculars perplexityText91%Zero-shot LLM scoring
Fast-Detect-GPTText90%Perturbation-based detection
XceptionNetImage95%Deepfake face + GAN detection
Error Level Analysis (ELA)Image89%JPEG compression forensics
Hive Moderation APIImage93%AI image generation classifier
RawNet2Audio96%Voice clone detection (ASVspoof)
Wav2Vec2 XLSRAudio94%Synthetic speech detection
MFCC-CNNAudio91%Spectral feature classifier
SyncNetVideo97%Lip-sync deepfake detection
RetinaFaceVideo95%Face detection + boundary check

TruthScan API Access

Ready to integrate TruthScan?

Free tier available. Enterprise plans include dedicated inference, SLA guarantees, webhooks, and custom model fine-tuning.

Request API Key →