Artists and illustrators
Artists may want to investigate work that appears to imitate their style, assess suspicious reposts or explain their own process when a piece is challenged. Keep source files, drafts, layer history and dated process material where practical. A detector can add one data point, but cannot by itself show that a work was copied or prove who made it.
Galleries, curators and marketplaces
Organizations reviewing many submissions can use detection as an initial signal for further review. Define what counts as prohibited AI use, how assistance is disclosed, how borderline cases are handled and how creators can appeal or provide context.
Automated scores should not become an unexplained gate. Record why a piece was flagged, restrict access to uploaded files and review false positives before taking action.
Teachers and competition judges
In education and competitions, a detector may help identify questions that deserve a conversation. It should not be the sole basis for a grade, accusation or disqualification. Explain expectations in advance and offer a way for students or entrants to show sketches, drafts, references and process notes.
Journalists and researchers
Researchers and publishers can use analysis to guide verification before publication or citation. Preserve the source or file, note whether it was resized or recompressed, and describe the detector and its limitations. Avoid reporting a result as fact when it is one probabilistic signal.
Choose the right review process
Before introducing a detector, decide who can submit files, what data is retained, who can view reports, how uncertainty is communicated and what action follows a result. The lower the tolerance for error, the more important independent evidence and human review become.
Read our accuracy guidance and privacy notice draft.