AI-Generated Code Detector
Signals, Not A VerdictDocumented stylometric signals + an independent LLM opinion — never a confidence score
Paste a code snippet
What this doesn't do: no reliable, published, general-purpose AI-vs-human code detector exists. Every signal here is individually weak and trivially fakeable in either direction — a careful human can write clean, well-documented code, and any LLM can be prompted to write messy code. This tool never outputs a probability or a "written by AI" verdict, and it shouldn't be treated as one even when several signals appear together.