Is Using AI in Job Interviews Cheating? Where Employers Draw the Line

Last updated: October 2026.

You use ChatGPT at work every day. You used it to tidy your CV, and last night you ran a mock interview with an AI interviewer. Now an ad promises an app that will whisper perfect answers during Thursday’s video interview, and a friend says “everyone does it”. Somewhere between those things there is a line. Where is it?

The honest answer is that it depends on when you use AI and what the employer has said. Most employers are happy for you to prepare with AI. Many accept AI help with application materials as long as everything is true. Almost none accept an AI feeding you answers while they are assessing you, unless they have said so. Here is how that spectrum works, what employers have actually published, and what you can safely use AI for.

Contents

The spectrum: prepare, apply, perform

Think of a hiring process in three stages. Your use of AI gets riskier the closer it gets to the moment someone is assessing you.

Preparation: broadly accepted

Using AI to get ready is no different in principle from reading a careers guide, asking a friend to quiz you or paying a coach. Asking an assistant to summarise a company’s annual report, generate likely questions for a data analyst role, or critique a practice answer is preparation. So is an AI mock interview that asks you questions out loud and tells you where you rambled. Every answer in the real interview still comes from you.

Employers that publish rules generally say so. Anthropic’s candidate guidance, last updated in July 2025, encourages applicants to use its Claude assistant to research the company, practise their answers, prepare questions and get suggestions on the clarity of their written answers.

Applications: usually fine, with conditions

Using AI to improve a CV or cover letter is common. A Gartner survey of job candidates, published in July 2025, found that 39% said they had used AI during the application process. The two conditions are honesty and the employer’s rules.

• Honesty: AI must not invent experience, skills or results. Greenhouse’s 2025 survey of 2,200 job seekers in the US, UK and Ireland found that 28% admitted using AI to generate fake work samples, and 32% had claimed AI skills they did not have. That is misrepresentation, whoever typed it.
• Rules: some employers ask for your own words. Anthropic asks candidates to “create your first draft yourself, then use Claude to refine it”, and says application answers about experiences you haven’t had are not allowed. Other employers ask you to tick a box confirming how you used AI.

Live interviews and tests: usually treated as cheating

An interview is a test of how you think, explain and solve problems. A tool that listens to the interviewer and puts an answer on your screen, or solves the coding task for you, replaces the thing being tested. Employers that have published rules treat it as misrepresentation unless they have explicitly allowed it. “Real-time interview assistants” are now a product category, and some vendors market their apps as “undetectable”. That is a claim about hiding the tool, not about whether using it is allowed.

What employers have actually said and done

These are the policies and reports we could verify in October 2026. Where the original is paywalled or internal, we name who reported it.

Amazon

In early 2025 Business Insider reported on Amazon’s internal guidance for recruiters. It says candidates should not use generative AI tools during interviews unless explicitly permitted, and that failing to follow this “may result in disqualification from the recruitment process”. The reasoning: AI help gives an unfair advantage and stops Amazon judging candidates’ “authentic” skills. An Amazon spokesperson said candidates must acknowledge they will not use unauthorised tools, where that applies.

Google, Cisco and McKinsey

In 2025 Google’s chief executive, Sundar Pichai, said the company now makes sure candidates have at least one face-to-face interview, “just to make sure the basics are there”. In August 2025 the Wall Street Journal reported that Cisco and McKinsey were also adding in-person stages, with recruiters citing candidates using AI tools off camera. The article quoted the head of a recruiting firm estimating that about a third of its corporate clients now asked for in-person interviews, up from around 5% the year before.

Anthropic

Anthropic, which makes the Claude AI assistant, publishes its rules for candidates at anthropic.com/candidate-ai-guidance. In short: use Claude to prepare and to refine drafts you wrote yourself, but no AI for take-home assessments or live interviews unless the company says otherwise. It is a useful model because it says clearly what is allowed, not just what isn’t.

What candidates think

Candidates don’t agree on where the line is. Greenhouse found that only 19% of US candidates said using AI in a live interview crosses the line, and 21% of Gen Z candidates thought AI was fair game in any circumstance. Employers’ published rules are much stricter than that. When the two disagree, the employer’s rules are the ones that decide your application.

How employers check

We won’t describe ways to get round any of this. At a high level, though, it helps to know that employers don’t rely on one trick:

• Trained interviewers. Amazon’s guidance reportedly tells interviewers to watch for candidates who type while being asked questions, seem to read their answers, or keep looking elsewhere.
• Follow-up questions. A polished first answer is easy to generate. Explaining why you chose that approach, or what you would change, is much harder to fake.
• In-person or supervised rounds for at least one stage, as Google and others now use.
• Written declarations. Many processes now ask you to confirm you won’t use unauthorised tools, which turns AI use into a clear breach of an agreement you signed.
• Identity and reference checks, which Gartner recommends as candidate fraud grows.

What happens if you’re caught

The consequences depend on when it comes out:

• During the process: you are removed from the process. Some employers keep notes, so reapplying later may be harder.
• After an offer: offers are usually conditional. Misrepresenting yourself can be grounds to withdraw one.
• After you start: if it later becomes clear you can’t do what you appeared to do in the interview, your employer may treat it as a performance problem or as dishonesty. Many employment contracts say false information in the hiring process can lead to dismissal.
• Your reputation: in 2025 a student who built a tool to cheat in coding interviews and filmed himself using it in an Amazon interview drew wide coverage. NBC News reported that Amazon complained to his university, which later suspended him over a related matter.

The quieter cost is the job itself. If an AI did the thinking in the interview, it will not sit next to you in your first team meeting.

What you can safely use AI for

Plenty. These uses help you perform better as yourself:

1. Research the company. Summarise annual reports, recent news and the job description, and ask what problems the team is likely to face.
2. Generate practice questions for the role, including technical ones. For a data role, ask for SQL, Excel or statistics questions at the right level.
3. Run mock interviews and ask for blunt feedback on structure, length and filler words. Our guide to the best AI mock interview tools compares the options, and we show a full session in how to practise a data analyst interview with AI.
4. Shape your stories. Paste in a rough account of a project and ask the AI to help you structure it (situation, task, action, result). Then rehearse it in your own words.
5. Polish your CV and cover letter after you have written the first draft, checking that every claim is true.
6. Prepare questions to ask them, and research salary ranges before you negotiate.

Some interview apps sell both: practice tools and a live “copilot”. The practice part can be useful. If you are considering one, read what it does with your recordings and CV first; our ParakeetAI review covers one popular example.

Frequently asked questions

Is it cheating to use ChatGPT to prepare for an interview?

No. Researching the company, practising likely questions and getting feedback on your answers with ChatGPT or any other assistant is preparation, the same as using a book or a coach. Employers that publish AI rules, such as Anthropic, explicitly encourage it.

Do I have to tell employers I used AI on my CV?

Only if they ask, and some do. Read the application form carefully: some employers ask you to confirm how you used AI, or to write the first draft yourself. Whatever you do, every claim must be true and you must be able to talk about it in the interview.

Is using AI during a video interview illegal?

Generally it isn’t a crime, but that is the wrong test. It usually breaks the employer’s rules, and if you signed a declaration that you would not use unauthorised tools, it is a clear breach. That can get you disqualified, or an offer withdrawn later.

Can employers tell if I’m using an AI interview assistant?

Often, yes, and not only through software. Interviewers are trained to notice read-out answers and unnatural pauses, follow-up questions expose gaps, and many companies now add an in-person round. Vendors’ “undetectable” claims are marketing, not a guarantee.

Is AI allowed in take-home tests?

Only if the employer says so. Some now allow it and want to see how you work with AI; others, including Anthropic, say take-homes should be done without AI unless stated. If the instructions are silent, ask.

What’s the safest way to use AI in a job search?

Use it before the interview, never instead of yourself during it. Research, practise, get feedback and polish your own drafts. Then walk in able to explain everything you’ve written and said.

Keep reading

Learn DataWhat Is Data Literacy and Why Does It Matter?10 min readLearn DataData-Driven Solutions for Social Impact3 min readAI for Data8 Perplexity Alternatives for AI Research With Sources (2026)11 min read