Privacy
The short version: your image is analysed in memory and discarded. It is not stored, not indexed, and not used to train anything.
Lifecycle of a submitted image
- 01
Sent with metadata intact
Your file is uploaded byte-for-byte. It is not re-encoded or stripped before it leaves your device, because the EXIF block and the file's original compression structure are themselves part of what the detector examines.
- 02
Held in memory only
The image is processed as a byte buffer for the duration of the request. It is never written to disk, never placed in a bucket, and never inserted into a database. There is no upload history to leak, because there is no store.
- 03
Sent to the classifier
Images are sent to a third-party inference provider for classification. The service receives the image bytes and returns label probabilities. We cannot guarantee how the provider handles data in transit. If that service is unreachable, the request falls back to local forensic analysis only.
- 04
Discarded
When the response is written, the buffer goes out of scope. Nothing about the image persists on our side.
What we do collect
- Technical request data — Your IP address and a timestamp, kept transiently to enforce the per-hour rate limit. The counters are held in the server's memory and reset on restart; they are not persisted or analysed.
- Aggregate operational metrics — Request counts, response times, and error rates, used to keep the service working. These are counts, not contents, and are not linked to images.
- Anonymous product analytics — Page views and which pages are reached, in aggregate, to decide what to build next. No account exists, so nothing is tied to an identity.
What we do not do
- We do not store, retain, or archive submitted images on our servers.
- We do not use submitted images to train or fine-tune any model.
- We do not sell or license submitted images or analysis results.
- We do not require an account, and there is no login to associate data with.
- We do not build a search index, gallery, or public feed from uploads.
Third-party data processing
Images are sent to a third-party inference provider for classification. We do not control how that provider processes data in transit or whether temporary copies are created during inference. If you are analyzing sensitive content, be aware that the image leaves our infrastructure. We recommend against uploading confidential, legally privileged, or personally sensitive images.
Your control over a submitted file
Because nothing is retained on our servers, there is nothing to delete on request — the image is already gone by the time you read the result. However, we cannot guarantee deletion from third-party systems.