Programming course submissions
Flag assignments worth discussing with a student, then ask them to walk through their code.
Estimate whether a code snippet was AI-generated and see which signals point that way.
AI coding assistants leave habits in the code they write. Comments that narrate every obvious line, very tidy but generic names, and a textbook structure that ignores the surrounding project are common examples. Human code tends to show shortcuts, uneven formatting and choices tied to a specific codebase. This detector reads your snippet and reports which of those signals it sees.
You get a rating of Likely human, Unclear or Likely AI-generated, plus the specific reasons. It is an estimate, not proof: short snippets, heavily edited AI output and very clean human code all blur the picture. Teachers, interviewers and code reviewers can use it to decide which questions to ask, never as the only basis for a decision.
Paste a code snippet or file into the box.
Press Analyze.
Read the rating and the signals, and use them to guide a conversation or closer review.
Flag assignments worth discussing with a student, then ask them to walk through their code.
Get talking points for a follow-up interview about how a candidate built their solution.
Spot generated code that may not fit project conventions and needs a closer look for correctness.
No. It gives an estimate based on patterns, and both false positives and false negatives happen. Never use it as the sole evidence for a grade, hiring or disciplinary decision.
Most common languages, including Python, JavaScript, TypeScript, Java, C, C++, C#, Go, PHP and Ruby.
A precise number would suggest more certainty than any detector can offer. A three-level rating with reasons is more honest.
Short snippets carry few signals, so the result is more likely to be Unclear. Longer code gives the tool more to work with.