Does Codility detect cheating?

You want to know whether something is watching you while you work. Some of it is, on every test, and more of it when the employer turns proctoring on. This section walks through the checks in the order they happen, so the timing is clear.

Every Codility Screen report includes a task timeline. As of September 2026, Codility's report guide documents two markers on it: when the candidate was "Out of focus" and where "Text pasted into the editor" appears. That holds whether or not the test is proctored, so a reviewer can see when you left the tab and where a block of code arrived all at once.

When the employer enables proctoring, those markers become tracked behavioral events:

• Tab switches: each time you move away from the Codility tab.
• Paste attempts: logged together with the code that was pasted.
• Copying the task: attempts to copy the task description.
• Speed: unusually fast completion of a task.

Some assessments also require full screen, a photo or camera check for identity, or the Codility App for Device Integrity, which Codility's documentation says can detect cheating tools running on the device. What none of these reveal is the content of another tab: Codility sees that you left, not what you looked at.

After you submit, the code goes through similarity comparison against a body of previous submissions and publicly available solutions. That check carries the most weight, because a well-known problem has thousands of prior answers to compare against.

Similarity comparison is not looking for identical text. It compares structure: the shape of the logic, variable patterns, the order of operations. Renaming variables and reformatting do not move the result much, because the underlying structure is what is being measured. The same approach also produces false alarms. A standard algorithm written the standard way looks like every other correct version of it, and there are only so many ways to write a binary search.

Codility also says it helps employers detect AI-generated code, but it does not publish how. The signals it does document are indirect: pasted text, time out of focus, unusually fast completion, and similarity to known solutions. Codility can also replay the order in which your solution was written, so a reviewer can see whether it grew line by line or appeared in one block.

Comparable breakdowns for other assessment vendors are collected on the detectability topic hub.

The report reaches a recruiter or an engineer with the flag attached, alongside your code, your timings, the focus and paste markers, and the edit history. What happens next depends entirely on the company. Some drop the candidate; many use it as a reason to add a live follow-up round where you walk through your own submission.

A data engineer applying to a logistics company was flagged for a solution that matched a common textbook implementation. The follow-up call took fifteen minutes: they explained the approach, adjusted it live when asked, and the process continued. Being able to defend your own code is what the flag ends up testing.

What SubcueAI records during a live interview, and what it stores afterwards, is set out on the security page.

SubcueAI assists live conversations. The native macOS and Windows app captures system audio plus your microphone and shows suggestions in a floating local overlay; the browser extension's Side Panel does live assist for a meeting tab on Chromium browsers, capturing only that tab's audio. No meeting bot joins the call, and nothing is injected into the meeting page.

A Codility assessment is not a conversation, so none of that applies. If the assessment is proctored or your screen is captured, an overlay is inside the capture, and that is a limit rather than a configuration problem. A test that requires the Codility App for Device Integrity is built to look for tools running on the device as well. Codility Interview, the live format where you code with an engineer, is closer to a conversation, but the interviewer sees the same focus and paste events as they happen.

The rounds SubcueAI is built for are the recruiter screen, the behavioral interviews, and the design discussion that usually follows a passed assessment. For the assessment itself, the useful preparation is timed practice, which runs on the mock interview page.