# The In-Interview AI Use Decision, Scored Round by Round

*You will be able to score any interview round to one of four verdicts - use openly, use and disclose, prep-only, or do not use - with a disclosure line ready to send.*

- Canonical URL: https://www.refolk.ai/candidates/guides/in-interview-ai-use-decision
- Pillar: Interviewing
- Format: Framework
- Published: 2026-09-05
- Last reviewed: 2026-09-05
- Reading time: 10 min

You are staring at a calendar invite for one interview round, and you need to decide one thing before you join: can you use an AI assistant during it, and if so, how do you say so without it reading as cheating. This guide is for candidates who want a per-round decision, not an abstract ethics debate. It hands you a scoring model, four verdicts, and a disclosure sentence you can send.

The reason this is hard is that the answer moved. Permission used to be a company property - Amazon bans it, some startup allows it. Now it is a round property. Google bans AI in its standard loops while piloting Gemini in one code-comprehension round. So the honest answer to "is using AI in interviews cheating" is: it depends on the round in front of you, and you can score that round in about twenty minutes.

## Why permission is now a round property, not a company one

Permission has migrated from the company to the round, which means a recruiter's company-level answer can be true and still wrong for your specific session. Employers now redesign scoring per round rather than per brand, so one loop can require an AI assistant while the next forbids it.

Look at the split. Meta introduced an AI-enabled CoderPad round in October 2025 for mid-level (E4) and senior (E5) Software Engineer roles, and pairs it with a separate AI-free round. Canva goes further and makes AI mandatory in Backend, Frontend, and Machine Learning engineering rounds. Google bans AI in standard loops but is reported to be piloting a Gemini-in-CoderPad format for code comprehension. Amazon bans it outright, with disqualification stated as the consequence. Goldman Sachs bans it in recorded video interviews.

| Employer | Verdict | Round scope |
|---|---|---|
| Meta | Allowed | AI-enabled CoderPad round, E4-E5, paired with one AI-free round |
| Canva | Required | Backend, Frontend, and ML technical rounds |
| Google | Split | Banned in standard loops; Gemini piloted in code-comprehension round |
| Amazon | Banned | Disqualification stated |
| Goldman Sachs | Banned | Banned in recorded video interviews |

Read the table twice. Meta and Google appear on the "allowed" side and the "banned" side of their own process. That is the whole problem in one grid. A candidate who hears "Meta is fine with AI" from a friend, and applies that to the AI-free half of the loop, has just made a company-level assumption about a round-level rule.

> **Rule:** Score the round, never the company
>
> A policy is only decidable at the round level. Get the specific round named in writing before you decide anything, because a true company-level answer can still be wrong for the session you are about to join.

The reason this migration matters for scoring is that it invalidates your cheapest source. You cannot ask "does this company allow AI" and act on the reply. You have to ask "does this round allow AI" and get the round named back to you.

## What the scored round is actually testing

An AI-permitting round is not testing whether you can type. It is testing AI fluency: prompting, verifying the output, and reasoning about it out loud. That is why underusing the approved tool can count against you as much as overusing a banned one.

The market signal here is loud. In 2026, according to Dice, 71% of tech job postings in the US require some form of AI fluency, a 181% year-over-year increase. When an employer builds a round where the tool is required, the round is a job proxy: they are watching how you drive the assistant, catch its mistakes, and explain your choices. Hand yourself the answer and stop thinking, and you fail the thing they are actually grading.

This skill is thinner on the ground than the hiring frenzy suggests, and it is geographically skewed. In Refolk's index of professional profiles, the exact skill these rounds test is underrepresented, which makes demonstrated fluency a genuine differentiator rather than table stakes.

| Segment | Count | Derived |
|---|---|---|
| US SWE, Prompt Engineering | 1,437 | baseline |
| India SWE, Prompt Engineering | 2,916 | 2.03x the US pool |
| US SWE, GitHub Copilot | 460 | 0.32x the Prompt-Engineering pool |

**0.32x - US software engineers listing "GitHub Copilot" vs those listing "Prompt Engineering"**

In Refolk's index, explicit tool-fluency listings are a fraction of the broader prompt-engineering pool, so a candidate who can demonstrate it in a round stands out.

The practical read: if a round allows AI, treat that as an invitation to show fluency, not a shortcut to skip the work. The verdict "use openly" is not permission to coast. It is a scored event.

## The four verdicts, and the signals that produce each

Every round resolves to one of four verdicts, and the verdict is a function of two things: how strong the permission signal is, and how exposed you are if you are wrong. Below is the map, then the scoring.

- **Use openly.** The round is AI-required or explicitly AI-allowed in writing, and the platform supports it. Drive the tool, narrate your reasoning, verify its output.
- **Use and disclose.** AI is plausibly fine but not explicitly named for this round, or you are on a platform where a detector could false-positive you. Use the tool and state that you are, in one plain line.
- **Prep-only.** No written permission, a standard live round, or a video screen where the norm is unassisted. Use AI to prepare a story bank and rehearse, then close it during the round.
- **Do not use.** A proctored lockdown, an explicit ban like Amazon's, or a recorded round like Goldman's HireVue. Any assistance breaks a rule you accepted, not just a norm.

The two axes that produce the verdict are permission strength and detector exposure.

#### The verdict matrix

Horizontal axis runs from Weak or silent permission to Strong written permission. Vertical axis runs from Low detector exposure to High detector exposure.

| Quadrant | What it means |
| --- | --- |
| Prep-only | No clear yes and little to gain live; prepare with AI, then close it. |
| Use and disclose | Permission plausible but a detector could misfire; use it and state so in one line. |
| Do not use | Silent or banned plus a lockdown or recording; any assistance breaks an accepted rule. |
| Use openly | Named permission and a platform that supports it; drive the tool and narrate. |

*Two axes decide it: how strong the written permission is, and how exposed you are to a detector or an accepted rule.*

Note where "do not use" sits: high detector exposure with weak or silent permission. A proctored assessment lands there by default. When you explicitly accepted full-screen lockdown and process monitoring, using assistance violates rules you agreed to, which is a different category of problem than breaking an unwritten norm.

> Permission strength decides whether you can; detector exposure decides how badly it goes if you are wrong.

## Read the detector surface before you decide

Detectors watch three surfaces, and knowing which one your platform uses tells you what "use and disclose" actually protects against. In-browser telemetry is the most common and cheapest bucket: tab switches, window focus loss, clipboard pastes, keystroke timing, mouse movement.

HackerRank runs tab-proctoring, copy-paste tracking on by default, and AI plagiarism detection cited at 93% accuracy on typing patterns. CodeSignal computes a "Suspicion Score" from paste events, typing-speed anomalies, and code-similarity checks. CoderPad logs pastes, plays back keystrokes, and can use the webcam, but it monitors the editor, not your whole desktop. Async video tools like HireVue read speech pace and hesitation and can log blur, but cannot see another screen.

| Platform | Primary signal | Documented limit |
|---|---|---|
| HackerRank | Tab and paste tracking, plagiarism AI (93% claimed) | Sees only its own page |
| CodeSignal | Suspicion Score: paste, typing, similarity | Assessment product only |
| CoderPad | Paste logging, keystroke playback, webcam | Monitors editor, not full desktop |
| HireVue (async) | Speech pace and hesitation, blur logging if enabled | Cannot see other screen content |

Two things to carry from this table. First, the webcam catches the off-screen glance but not the on-screen read: it cannot tell whether you are reading the question or reading an overlay positioned next to it. The tell is the glance, not the read. Second, and more important, none of these is the real detector.

The most reliable detector is conversational. It is one unscripted follow-up about code you did not write. An overlay defeats tab and screenshot telemetry, but it cannot help you explain, in your own words, a design decision you never made. Reading a generated paragraph produces every signal at once: uniform latency, scanning eyes, documentation-style phrasing, and zero depth on the follow-up.

> **Watch out:** The overlay defeats the wrong detector
>
> Undetectable-overlay claims target software telemetry, which is not the load-bearing risk. The risk is the follow-up question, and no overlay can defend code you cannot explain. If your verdict depends on the tool being invisible, your verdict is wrong.

CoderPad is the one platform with a documented per-pad toggle: an interviewer enables or disables AI Assist for an individual pad from Pad Settings, and the organization default does not gate availability. This is the surface Meta uses and Google is reported to be piloting. If your round runs on CoderPad and the interviewer says AI is on, that is a strong, verifiable permission signal.

## The procedure, run in under thirty minutes

Run these seven steps in order before you join the call. Each has a clear "done" state, and the whole sequence fits in the gap between scheduling and the interview.

#### Score one round to a verdict

1. **Get the written policy** - Email the recruiter and ask whether the round is closed-book, open-book, or AI-allowed. Done when you have a dated reply that names the specific round and the permitted tool.
2. **Identify the round type** - Classify as AI-required, standard live coding, behavioral, take-home, or proctored assessment. Take-homes invert the rule: AI is usually assumed and rarely policed because the bar is what you ship and can defend.
3. **Identify the platform and detector surface** - Determine whether it runs on CoderPad, HackerRank, CodeSignal, async video, or a proctored lockdown. Done when you know what the detector watches and what it cannot see.
4. **Run the out-loud self-test** - Ask whether you would be comfortable saying your planned AI use aloud. Prepping with a coach passes; reading answers off a hidden overlay fails. Done on a clean pass or fail.
5. **Score the round and land a verdict** - Weigh permission strength against detector exposure and pick one of use openly, use and disclose, prep-only, or do not use. Done when one verdict is chosen for this round.
6. **Draft the disclosure line** - Write one plain, boring sentence that states your tool use without a speech. Recruiters want a credible process, not an essay. Done when you have one sendable sentence.
7. **Prep the defense** - Rehearse explaining any AI-produced output in your own words, because the real detector is the follow-up. Done when you can defend every line.

Step one carries the most weight, so use a script that reads as process-minded rather than nervous. Asking costs nothing and carries no stigma. Assuming is the expensive option in both directions: assume permission and you risk disqualification, assume prohibition and you underperform.

```template
title: The pre-round permission ask
note: Send to the recruiter after scheduling. It reads as a candidate who thinks about process, and the reply removes the only version that ends badly.
Hi [name], quick logistics question before the [round name] round.

Are any of the rounds designed for me to use my normal tools, or should I plan to work unassisted throughout? Happy either way - I just want to prepare correctly.

Thanks.

## Frequently asked questions

### Is using AI in interviews cheating?

It depends entirely on the round, not on a general principle. In an AI-required round at an employer like Canva, not using the tool can count against you; in a proctored lockdown you explicitly accepted rules that forbid it, so any assistance breaks an agreement. The cheating line is drawn by the specific round's stated policy, which is why you get it in writing before you decide.

### Can I use ChatGPT during an interview?

Only where the round permits it, and you should confirm that in email first. Some technical rounds now require an AI assistant, others ban it, and permission increasingly varies by round within the same company. If the policy is silent, run the out-loud self-test: if you would not be comfortable saying you used it, treat the round as prep-only and keep the tool closed.

### How do companies detect AI use in coding interviews?

They watch three surfaces: in-browser telemetry like tab switches and pastes, webcam behavior like off-screen glances, and conversation. HackerRank claims 93% plagiarism accuracy on typing patterns and CodeSignal runs a Suspicion Score, but the most reliable detector is a single unscripted follow-up about code you did not write. Overlays defeat the telemetry, not the follow-up.

### How do I disclose AI use in an interview without it reading as cheating?

Use one plain sentence, sent before or at the start of the round, that names the tool and the task. Recruiters want a credible process, not a speech, so the safest tone is boring. A disclosure line also protects you against false positives: detectors misclassify non-native English writing as machine-generated 61% of the time, and a stated disclosure defuses a wrongful flag.

### What are the flag rates for AI use in interviews, and are they reliable?

One widely cited figure is 38.5% of candidates flagged across 19,368 live interviews between July 2025 and January 2026, rising to 48% in technical roles. Treat these cautiously: the data comes from a vendor that also sells detection, recruiter-observed use is reported at 35%, and detectors misfire on non-native English. The spread between sources is the real finding, not any single number.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/candidates/guides/in-interview-ai-use-decision*