ART SHIELD

Beta

Artifact IV

Hides a faint copyright notice inside your image, readable by machines but meant to be invisible to people.

Beta, and an experiment, not protection. In early tests ChatGPT and Gemini read past the hidden notice and answered normally, and it does nothing against offline AI training. No account needed; uploads and results are deleted automatically after 7 days.

Original

Drop or click to upload
PNG / JPG · up to 4 megapixels

Mode

Strength

±2

Harder to see, easier to missEasier to see, easier to read

Above about ±6 the text may start to show on flat areas. Try several strengths to find where models react.

Directive

Protected

Result appears here

Publish the PNG at its native size. Instagram, X and other platforms recompress images and can erase the signal. Some models ignore text found in images, and this does not stop offline LoRA training. It is one layer, not a guarantee. Your original is kept privately as your master copy.