Does CodeGrade Detect AI Code?

What CodeGrade's own help center says about its plagiarism checker, AutoTest and AI Assistant, what your instructor can see, and why none of it is described as detecting AI-written code.

0 / 8,000 characters

CodeGrade is an autograding platform used in many programming courses, often inside Canvas, Moodle, Blackboard or Brightspace. If your assignments go through it, you may be worried that it can spot code written with ChatGPT or another assistant. The short answer is that CodeGrade's documentation does not describe AI code detection. It describes a plagiarism checker that compares code, an autograder called AutoTest, and an optional AI Assistant for students. Everything below about CodeGrade comes from its official help center and product pages, which we read in October 2026; check the linked pages for the current wording.

What CodeGrade's plagiarism checker does

CodeGrade's help article on detecting plagiarism says the checker compares each student's submission with each other and flags cases where students' code is identical. It says the comparison works even when students have tried to hide copying by reordering code structures or renaming variables. An instructor sets the language and a minimum similarity threshold (the default is an average of 50%), and can optionally:

  • include the submissions from an older assignment,
  • upload an archive of other submissions to compare against, and
  • upload base code so that shared starter code is excluded.

The resulting report shows suspected pairs side by side, with matching sections highlighted, and it can be exported. CodeGrade's plagiarism checker product page adds that it uses JPlag, an open-source similarity engine, and compares within the current class, across previous semesters and against external sources that the instructor uploads.

Neither page mentions AI-generated code. The checker is about matching code against other code.

Why that is not AI detection

A language model writes a fresh answer each time, so an AI-written solution usually does not match any single existing program. A similarity checker can show that two submissions share code, but it cannot say whether that code came from a classmate, a website or a model. The one case where AI use can surface is when several students submit closely matching generated code for the same task; then the overlap shows up like any other match. For more on how similarity tools behave, see does MOSS detect AI code.

CodeGrade itself has written about this. A CodeGrade blog post from March 2023 said that ChatGPT detectors, classifiers and plagiarism scanners are not reliable, and that it is implausible they will ever be for generated code. That is an opinion piece rather than product documentation, and it is a few years old, but it is consistent with CodeGrade not offering an AI detector.

What AutoTest checks

CodeGrade's student getting-started guide explains that you can hand in work by writing it in the online editor (if your instructor enables it), uploading files, or connecting a Git repository so that each git push uploads your work. If your instructor sets up automatic tests, you get instant feedback and can click each test to see whether it passed and why. AutoTest checks what your instructor configured it to check. The documentation does not describe it judging who or what wrote the code.

The AI Assistant, and what instructors can see

CodeGrade does use AI, but as a helper. Its help page on the AI Assistant says instructors can set up assistants for an assignment, in three types: a Learning assistant focused on understanding concepts, a Feedback assistant that comments on code quality, and a Coding assistant that can generate code and solutions. Instructors can configure content guardrails to control what the assistant responds to. The page says the feature is in Beta.

One line on that page matters a lot for students: both students and instructors can view and review all conversations with the assistant. So if your course turns it on, your chats are visible to your instructor. That is not AI detection, but it does mean the help you asked for is on the record.

What your instructor can realistically see

  • Your submitted code and your AutoTest results.
  • Similarity matches with classmates, earlier semesters or uploaded code, if they run the plagiarism checker.
  • Your AI Assistant conversations, if the assistant is enabled for the assignment.
  • How your work compares with your earlier work, simply because they read it.

None of these is an automatic "AI" label. Instructors who are concerned about AI use usually rely on context: a sudden change in style, code that uses things the course has not covered, or a student who cannot explain their own solution. Our guide on whether professors can tell you used ChatGPT covers those checks in more detail.

What this means for you

  • Read your course's AI policy. Some courses allow the CodeGrade assistant, or even require AI exercises. Others forbid outside tools. The rules come from your instructor, not from the platform.
  • Keep your history. If you submit through Git, commit as you go. Drafts and commits are the clearest evidence of how your solution grew.
  • Understand what you hand in. Be ready to explain any line, change it on request, or extend it. That answers a similarity question and an AI question at the same time.
  • Disclose help you were allowed to use. If AI assistance was permitted, saying so up front avoids misunderstandings.

If you are unsure whether something counts as allowed help, ask before the deadline rather than after. A short message to your instructor or TA, such as asking whether you may use an assistant to explain an error message, costs nothing and leaves a written answer you can point to later.

Checking how your code reads

You cannot run CodeGrade's class-wide checker yourself, but you can see whether your code has the style patterns people associate with AI output. Paste it into the detector at the top of this page, read the reasons it lists, and compare them with how you really write. Use it to understand what a grader might ask about, not as a verdict, and be ready to talk through your solution.

Read more

Frequently asked questions

Does CodeGrade detect AI-generated code?

CodeGrade's documentation does not describe AI detection. Its plagiarism checker compares students' submissions with each other and with older or uploaded code. Its AI features are an assistant for students and AI-based exercises, not a detector. A 2023 CodeGrade blog post even argued that ChatGPT detectors are not reliable for generated code.

What does CodeGrade's plagiarism checker compare my code with?

According to CodeGrade's help center, it compares each student's submission with every other submission in the assignment. Instructors can also include an older assignment or an uploaded archive of submissions, and exclude base code. CodeGrade's product page says it uses JPlag and can compare across semesters and against external sources that are uploaded.

Can my instructor see my chats with the CodeGrade AI Assistant?

If your course uses it, yes. CodeGrade's help page says both students and instructors can view and review all conversations with the assistant. Treat those chats as part of your coursework record.

Can CodeGrade catch code I got from ChatGPT?

Not directly, based on its documentation. The plagiarism checker finds code that matches other code it was given. If classmates submit closely matching AI output, the overlap can show up as similarity, but the tool does not say where code came from. Instructors may still notice through other means, such as asking you to explain your work.

My CodeGrade plagiarism report flagged me. What now?

Ask what was matched and against whom. Short or standard tasks often look alike. Gather your drafts and Git history, explain how you built your solution, and offer to walk through the code. Check your course's AI policy so you know what was allowed.

Try the detector

Have a snippet you are unsure about? Paste it into the free AI Code Detector to get a verdict, an AI probability and the reasons behind it. No sign-up, and your code is not stored.