Which companies ban AI in interviews (2026)

Some employers now disqualify candidates for using AI during the interview. A handful build the interview around AI and expect you to use it. Most have said nothing in public. Amazon and Anthropic tell candidates to keep AI out of the live round or risk being cut, while Canva and Shopify hand you the opposite instruction and want to watch you drive the tool. Below is the dated, sourced picture as of July 2026, company by company, and the one move that wins in every column is being good enough that you would not reach for AI in the room even if you could.

By UnchartedCareer

Last updated: July 2026

Which companies ban, allow, or stay silent on AI in interviews?#

No official registry of this exists, so here is a maintained one, built only from dated sources you can check yourself. Each row is a named employer's stance on candidates using AI during the interview itself, not on the company using AI to screen you.

| Employer | Stance | What they tell candidates | Source | As of | |—-|—-|—-|—-|—-| | Anthropic | Bans in the room | Live interviews are "all you," with no AI assistance unless Anthropic says otherwise. Take-home assessments should be done without Claude unless a task allows it. | Anthropic candidate AI-usage guidance | July 10, 2025 | | Amazon | Bans in the room | Recruiter guidance tells candidates not to use GenAI tools during the interview unless explicitly permitted, and warns that failing to comply "may result in disqualification." | Amazon internal guidance, reviewed by Business Insider (via IT Pro, GeekWire) | March 14, 2025 | | Google | No published ban | No candidate-facing AI ban in writing. CEO Sundar Pichai said Google will add at least one in-person interview round, to confirm a candidate's fundamentals are solid in an AI era. | Lex Fridman Podcast, episode 471 (Pichai) | June 5, 2025 | | Canva | Requires AI | Expects Backend, Machine Learning, and Frontend engineering candidates to use AI tools like Copilot, Cursor, and Claude in technical interviews, and tells them so in advance. | Canva Engineering Blog | June 11, 2025 | | Shopify | Allows any AI | Encourages candidates to use whatever AI tools they want in coding rounds. Its VP of engineering says candidates without a copilot "get creamed by someone who does." | The Pragmatic Engineer, interview with Farhan Thawar | July 2, 2025 | | McKinsey | Requires its AI (pilot) | Asked some graduate candidates to use its internal AI, Lilli, during case interviews, and assessed how they prompted it and judged its output. | CFO.com | January 2026 |

Read the "what they tell candidates" column, not the stance label alone. Amazon's line is about GenAI "during your interview," and Anthropic's is about the live round and the take-home, so both still leave AI open to you while you prepare. The allow side is narrower than it looks too. Canva's instruction covers engineering roles, and Shopify's is about coding rounds where the job itself gets done with a copilot.

How current is this, and what are its limits?#

Treat it as a starting map, not gospel. Policies change, and they vary by role and by interview stage inside the same company. A stance that held in 2025 can flip in a quarter, which is exactly what Anthropic did when it went from a blanket AI ban to encouraging AI on applications while keeping the live-interview ban intact (Fortune, July 21, 2025). Everything above was last verified July 2026 against the dated sources named in the table. The employer's own written instructions override any of it, because the invite in your inbox is the rule and this list is only the guide.

What if your employer is not on the list?#

Assume the interview is AI-off until the invite says otherwise. That is Anthropic's own default, spelled out to candidates. Treat AI as prohibited in the assessment unless the company tells you it is allowed (Anthropic candidate AI-usage guidance, July 10, 2025). Most employers have published nothing, and silence is not permission.

So before your next round, do the boring homework. Read the invite and any assessment instructions line by line for the words "AI," "generative," or "assistance." If it is silent, email the recruiter and ask plainly whether AI tools are allowed in the interview. If you still cannot get a clear yes, treat it as a no. The downside is lopsided. Nothing is lost by skipping a tool you were allowed to use, and an offer can vanish for using one you were not. If you are still weighing whether to touch AI at all, our guide on whether it is safe to use AI in an interview walks that line case by case.

Why do the AI companies land on opposite sides?#

Because they are testing different things, and the stance follows the test. Anthropic sells the model and still bars it from the live interview, because it wants to watch you think through a problem in real time, unaided (Anthropic, July 10, 2025). Canva and Shopify made the copilot part of the job, so they made it part of the interview. Shopify's VP of engineering, Farhan Thawar, put it bluntly in a July 2025 interview with The Pragmatic Engineer: let candidates use whatever they want, because the ones who skip a copilot "get creamed by someone who does."

The split tracks what each room is trying to measure. A behavioral or live problem-solving round measures whether you can reason and defend a decision out loud, and a copilot reading you answers wrecks that signal. An AI-native coding round measures whether you can steer the tool and catch its mistakes before they ship, which you also cannot fake without the underlying skill. Both reward the same thing. You, competent, under a real question.

What is the move that works in every column?#

Be good enough that AI in the room would not change your answer. That preparation survives whichever column your employer sits in, and it holds even where AI is welcome, because a copilot amplifies a strong candidate and exposes a weak one. Where AI is banned, the skill is all you have. Where AI is required, the skill is what lets you catch the model's bad suggestion instead of shipping it.

You build it the slow way, and you can start tonight with no tool at all. Open your laptop camera and answer one hard question out loud with no notes, something like "tell me about a time you owned something that broke and what you did." Watch it back and find the three tells that show up in any interview: where you reached for filler, where your pace ran away, and the moment your eyes left the lens. Redo it until the follow-up stops rattling you. If your next round is a recorded video screen instead, how to pass a HireVue interview covers that exact format.

A recording teaches you your tells. It will not ask the question you did not see coming, and the unseen follow-up is where most answers fall apart. When you want that follow-up thrown at you on demand, scored and repeatable, AI interview practice is built to do it, after you have run the manual drill yourself.