How Protect works

Protect marks your image as not for AI training. It is a label, not a lock.

What it does

• Saves a copy next to your file, named "-protected". Your original is never changed.
• Writes into the copy, in the photo industry's standard fields: Data Mining "Prohibited for AI/ML training" (or "Allowed"), your copyright notice, Copyright Status "Copyrighted", and usage terms.
• Reads the copy back to confirm, then records its fingerprint and the time in your ledger on this device.

What it doesn't do

• It doesn't stop copying. A screenshot, a re-save or most social sites remove the notice. Your copyright still exists, but the label is gone.
• It doesn't prove you made the image. The ledger shows which file you registered and when.
• It doesn't change the pixels or add a watermark.

Where it counts

• EU: copyright law lets owners opt out of AI training in a machine-readable way, and makers of general-purpose AI models must have a policy to respect it. Courts haven't settled whether a notice inside the photo is enough on its own.
• US: there's no opt-out right for AI training, but removing a copyright notice on purpose to hide infringement is unlawful.

The AI check

Bitphoto reads the evidence in the file: signed Content Credentials (a verified signature) and AI or camera tags (a lead, shown as "Not proven"). It doesn't analyse pixels, so a screenshot or a stripped file shows "Unknown".

Your part

Only protect images you own or may mark. Keep your original camera files. The ledger shows which file you registered and when, not who made it; your originals help show that.