from zerobyte import Client
client = Client(api_key="0b_key_...")
# Detect if an image is AI-generated
result = client.analyze(open("image.png", "rb").read())
print(result.verdict) # "ai_generated"
print(result.confidence) # 0.94Neural classifier trained across DALL·E, Midjourney, Stable Diffusion, and Flux. Returns a probability score with every call.
Ed25519-signed proofs anchored in a Merkle transparency log. Any modification breaks the signature — tampering is mathematically impossible to hide.
Perceptual fingerprinting means verification works after screenshots, re-encoding, cropping, and compression. No metadata to strip.
Every stamped asset is indexed in an append-only registry. When content matches, you get the full provenance chain — provider, model, timestamp.
POST an image to /v1/analyze. Get back AI probability, origin proof, and confidence score. No ML expertise required.
Stock photo sites and design platforms are drowning in AI uploads sold as original work. 0byte detects them at the upload gate — before they reach your catalog.
Verify whether an image is AI-generated before it runs. When it's in the registry, trace it to the exact model and timestamp. When it's not, the neural classifier still flags synthetic content.
Detect AI-generated content at scale for platform moderation. Flag deepfakes, synthetic media, and AI-generated profiles. One API endpoint, your existing pipeline.
Stamp every generation at creation. Give your users provenance they can verify. Ship with built-in content authenticity — differentiate on trust.