How To Detect AI Art In Student Assignments

When a student hands in artwork that looks too polished, or oddly generic, it's tempting to run it through a detector and act on the score. That's risky: visual AI detectors make mistakes, especially on digital art, and a false accusation damages trust. Process evidence is far more reliable.

Ask for the process

• Sketches and drafts showing how the piece developed.
• Working files with layers, such as Procreate, Photoshop or Krita files.
• Time-lapse recordings, which Procreate and some other apps create automatically.
• Reference images the student used.

Making these part of the assignment from the start avoids awkward conversations later.

Have a conversation

Ask the student to talk through choices: why this color, how they drew the hands, what they'd change. Someone who made the work can usually explain it. Approach it as curiosity, not an accusation.

Check the file

If the student submitted an original file, its metadata may show the app used. AI Image Detector (made by Patrick Bushe, who runs this site) reads metadata in Chrome and flags AI generator names, Stable Diffusion settings and AI tags; it also shows editing software such as Photoshop. Many school platforms re-save uploads and strip metadata, so check the file the student sent directly, and treat "No Metadata Found" as inconclusive.

Visual clues

• Detail that's uneven: highly rendered in places, mushy in others.
• Hands, text or symbols that don't make sense.
• A style very different from the student's earlier work.

None of these prove anything on their own; students experiment.

Set a clear policy

Say what's allowed: AI for brainstorming, for reference images, or not at all. Students can't follow rules they haven't been given. Tools such as Turnitin focus on AI-written text, not images, so plan around process rather than detection.

See AI image detectors for teachers and a teacher's guide to AI images.