Why we are building this
Generative tools have changed how visual work is made, shared and evaluated. Artists may want to understand whether a piece attributed to them was generated or altered. Galleries, marketplaces, teachers and researchers may need a consistent way to review submissions. A useful checker should make that investigation easier while respecting the limits of automated analysis.
AI Artwork Checker is designed around one principle: show the evidence behind an assessment. A single score can hide uncertainty. A report should explain which signals were observed, how strongly they point in one direction and where the result remains inconclusive.
Made for artwork
Artwork spans paintings, illustrations, digital compositions and mixed media. A detector should account for that range. It should not treat every file as a photograph, or assume a particular style or subject proves how the work was made. Our examples focus on art, including fully generated work and pieces partially edited or enhanced with generative tools.
Evidence, with context
Pixels, file metadata and known generator patterns can each contribute clues. None is a perfect record of authorship. Metadata can be stripped, pixels changed by compression, and new models can produce unfamiliar traces. A report should help a person investigate, not replace their judgment.
We aim to make the method understandable and to be clear when evidence is weak or conflicting.
Who it can help
Artists can review suspicious reproductions or document their own work. Curators and marketplaces can add a screening step. Teachers and judges can use a structured report as one part of a broader review. Researchers and journalists can make verification more transparent.
Results should be considered alongside provenance, process documentation and other relevant context.
What is live today
This website describes the intended product experience. Some examples, usage figures, service guarantees and features may be illustrative until the corresponding checker service is available. We will distinguish product capabilities from examples as implementation details are finalized.
Read how accuracy and limitations are handled, explore the analysis process, or contact the team.