No, unless you are told otherwise. Internal guidance reported by Business Insider in early 2025 tells candidates not to use generative AI tools during interviews unless explicitly permitted, and warns that breaking the rule may lead to disqualification. Recruiters are told to watch for typing during questions, answers that sound read aloud and eyes that wander off screen.
What exactly does the guidance say?
The rule, as ITPro reported it in March 2025 from Business Insider's coverage of internal documents, is that candidates should not use generative AI tools in an interview unless that use is explicitly permitted. The consequence is stated plainly in the guidance: “Failure to adhere to these guidelines may result in disqualification from the recruitment process.”
We found no public Amazon page with this wording, so the rule is known from press reports of internal material. That makes your recruiter's instructions the final word for your own loop.
Why does Amazon ban it?
The guidance describes real time AI help as an unfair advantage that stops Amazon from assessing a candidate's own skill. The report links it to a video in which an AI company showed its coding tool being used during an Amazon interview, which raised concern inside the company.
The reasoning is not unique to Amazon. In an interviewing.io experiment, candidates told to use ChatGPT secretly passed 73% of interviews built on verbatim LeetCode questions, against 53% for a control group, and no interviewer raised a concern about cheating. Custom questions held up much better, with a 25% pass rate.
What do Amazon interviewers watch for?
The reported guidance lists three signs: typing while a question is being asked, answers that sound read rather than spoken, and eyes that wander away from the camera. None of these proves anything on its own, but each invites follow up questions, and follow up questions are where hidden help shows.
interviewing.io's 2025 survey adds that half of the Amazon interviewers who answered expected in person interviews to return, though one noted that Amazon had not yet addressed the question internally.
What if a recruiter permits AI for one exercise?
Then that exercise allows it, and only that one. Ask which tools are permitted, whether you may use them for the whole exercise or part of it, and whether the interviewer will watch. Keep the permission in writing, such as the recruiter's email, so there is no doubt later.
How should you prepare for an Amazon loop?
Prepare to work unaided. Practise coding problems under a timer with no assistant, talking through your reasoning as you go, because explaining your choices is what interviewers can see. Practise on custom problems as well as familiar ones, since the interviewing.io experiment suggests that is where interviewers move when they worry about hidden help.
Practising with AI is still useful between interviews. Using an assistant to review your solution after you finish, or to generate variations of a problem, is preparation, not cheating. What the guidance rules out is help during the interview itself.
If your target list mixes companies, our company policies page shows which ones allow AI in which rounds, with the source and date for each.
Do other large employers take the same line?
Several do, by default. Anthropic asks candidates to use no AI in take homes or live interviews unless it says otherwise, and Deloitte UK bans AI in its online assessments and interviews. Cursor allows no AI beyond autocomplete in its first technical screens. Meta and Google are the exceptions among the largest companies, and only for one round each: Meta's AI enabled onsite round and Google's pilot. Our list of who allows AI has the detail.
Does the ban apply to online assessments too?
The reported guidance is about interviews. For an online assessment, follow the instructions that come with it, and if they say nothing about AI, ask before you start. Assessment platforms used by many employers record sessions, and a vendor such as CodeSignal reports that it flags a rising share of proctored sessions for suspected cheating, so silence is not permission.