Tier 1 local • Tier 2 optional

Check an image for AI provenance

Find explicit AI generator and provenance signals, with a likelihood score tied to evidence in the file.

Tier 1 stays local
Tier 2 requires consent
No account

Check locally first

Metadata evidence stays on your device.

Analyze pixels optionally

Upload only after clear consent.

Read results carefully

A likelihood score is an estimate, not proof.

Short answer: AI Remover Pro Tier 1 checks C2PA Content Credentials, generation parameters, workflow data and other supported AI metadata locally in your browser, with no image upload. A positive result is useful evidence. A negative result is inconclusive because metadata can be removed and pixel-level signals such as SynthID are not inspected — an optional Tier 2 can analyze pixels instead, only after you explicitly consent to that upload.

What this AI detector can prove

A named generator in Content Credentials or an embedded generation workflow is strong provenance evidence. A generic marker is reported separately because it may describe editing rather than prove that every pixel was generated.

What the AI likelihood percentage means

Tier 1 shows a high likelihood only when the file contains explicit generator or workflow evidence. The percentage describes confidence from file evidence; it is not a pixel-classifier result. When no marker is found, the answer stays inconclusive instead of pretending the image is 0% AI.

Private by design

Tier 1 reads the file in this browser tab. It makes no image upload request and stores no filename, thumbnail or image bytes. Close the tab and the selected image is gone.

How to check an image for AI provenance

Choose or drop a JPEG, PNG, WebP or HEIC image.
Read the evidence-based result for each file.
Open the metadata viewer for technical detail or remove supported AI metadata in a separate action.

Frequently asked questions

Can this prove that an image is human-made?

No. Missing metadata does not authenticate an image. It may have been stripped during export, editing or posting.

What counts as strong AI provenance?

A named AI generator in Content Credentials or embedded generation parameters and workflow data are treated as strong evidence.

Can it detect Google SynthID?

Not in Tier 1. SynthID is a pixel-level watermark, while this private checker inspects metadata and provenance records.

Does checking change my photo?

No. The detector is read-only and does not create a modified copy.

How is the AI likelihood calculated?

Tier 1 maps explicit provenance evidence to a cautious score: a named generator or generation workflow is high-confidence, a generic marker is medium-confidence, and missing metadata remains inconclusive.

Are my images uploaded?

Tier 1 stays in your browser and makes no image upload. Tier 2 is optional, remains locked until you consent, and uploads only the selected image for pixel analysis. AIRemover.Pro keeps no image copy.

Is the Tier 2 percentage proof?

No. It is a model estimate based on image pixels. Compression, editing, screenshots and unusual camera images can cause false positives or false negatives, so use it as one signal alongside provenance and context.

How is this different from the Content Credentials checker?

This detector reads C2PA/provenance markers and can optionally score pixels for AI likelihood (Tier 2). The Content Credentials checker only reads C2PA and other metadata — no pixel analysis, no Tier 2 upload. Use this page when you want a likelihood score; use the checker when you only want to read what metadata is present.