Seven AI detectors.
One method.
Photos, texts, voices, videos, documents, code and websites. Each analysis checks provenance first, runs an open-source detector, then explains the verdict in plain words. 7 detectors are open in beta; the others are coming soon.
AI Content Detection
Photos, videos, voices, texts, documents, code and websites: each analysis checks provenance first, runs an open-source detector, then explains why.
Generated images, cloned voices and AI-written texts now look real. A clear, explained verdict helps you decide before you trust, share or act.
A verdict you can read,
not a bare percentage.
Every analysis ends with a verdict in three levels and a short explanation of what was found. A result is an indication, not proof: detectors can be wrong, especially on compressed, edited or very short content.
Provenance first
We read content credentials (C2PA) and metadata. When they clearly say how the content was made, that settles it.
An open-source detector
If provenance is missing or inconclusive, a specialised open-source model scores the content — passage by passage, segment by segment or frame by frame.
A second opinion, then an explanation
In the grey zone, Claude (an LLM) gives a second opinion. Every verdict comes with a plain-language explanation and the limits of the method.
Start free.
10 free credits every month. No credit card required.