What the AI code detector finds
AI help usually leaves marks in the code or around it. The detector reads the characters you paste and reports three kinds of evidence, each with the matching text:
- Declared AI use. Comments and notes such as “generated by GitHub Copilot” or “written with ChatGPT”.
- AI tool mentions. Names such as Copilot, ChatGPT, Claude, Gemini or Cursor, including attribution lines like a commit trailer naming Claude.
- AI libraries. Imports of AI service SDKs, for example an OpenAI or Anthropic client, which show the code talks to a model.
The report keeps these apart. A comment saying a function was generated is a different claim from code that calls an AI service, and both differ from a passing mention of a product name.
Evidence plus an AI-likelihood score
Every declared AI mark comes with the exact line it was found on, and the snippet gets an AI-likelihood score from isGenAI’s code detector. Together they give you a finding you can quote in a review, a classroom or an audit, and a score that tells you where to look first.
How to check code for AI use
- Copy the file, function, diff or README you want to review, including its comments.
- Paste it into the checker above and select Check my code.
- Read each finding and the exact line it quotes, then open “How we checked” for the method.
- Compare the findings with the commit history, the pull request and the author's own account, then save the report with your notes.
Checking commits, pull requests and repositories
Commit messages and pull request descriptions often carry the clearest marks, such as a co-author trailer or a note that a change was drafted with an assistant. Paste them alongside the code. To check many files automatically, call the same reader from your own scripts with the free isGenAI API, which takes code with "kind": "code" and needs no key.
Checking prose instead? Use the AI text detector. Have a picture? Use the AI image detector. The full scope is in the checker methodology.