AI Code Detector
Spot assistant-style code with line-level evidence
Code generated by assistants leaves recognisable traces: narrated step-by-step comments, template placeholders, pasted Markdown fences and uniform documentation. AI Inspector Lab flags them with exact line numbers and, when a code model is configured, adds a per-file likelihood.
What the inspector checks
Line-referenced findings
Every heuristic finding links to a file and line range so reviewers can jump straight to it.
Explicit markers
Comments such as “Generated by ChatGPT”, leftover ``` fences and “YOUR_API_KEY_HERE” placeholders are surfaced first.
File and directory breakdown
Archives are summarised per file and per directory, excluding dependencies, lockfiles, minified and generated code.
Safe by design
Uploaded code is treated as text. It is never executed, compiled, installed or imported.
Who uses it
- Hiring take-home assessments
- Code review of third-party contributions
- Course assignment review
- Open-source licence and provenance audits
AI Code Detector FAQ
Which languages are supported?
Any text source file is analysed; language-aware comment handling covers more than 40 languages including JavaScript, TypeScript, Python, Rust, Java, C#, Go and PHP.
Is uploaded code ever run?
No. Archives are unpacked with strict size, ratio and path checks, and files are read as plain text.
Do comments matter?
Narration-style comments are one of the clearest assistant patterns, which is why we report them with line references.
Other detectors
Get the AI Code Detector with Pro
Text, document, image and code inspections are included with Pro and Team - private by default, with Compare and the REST API.