Usually not. The exception is a pilot reported in May 2026: from the second half of 2026, junior and mid level candidates on select US teams get a code comprehension round in which they use Gemini, Google's approved assistant, to read, debug and optimise existing code. Elsewhere Google has moved part of its loop back in person, which rules out hidden help.
What is the Google pilot?
Business Insider reported the pilot from an internal Google document, and the report was republished by Moneywise on Yahoo Finance on 14 May 2026. In the new code comprehension round candidates are asked to read, debug and optimise an existing body of code with an approved assistant. A Google spokesperson confirmed that the assistant is Gemini.
According to the report, interviewers evaluate AI fluency, described as prompting, validating the output and debugging. Google's vice president of recruiting, Brian Ong, said the pilot is meant to reflect how Google's teams now work.
Which candidates are in the pilot?
Junior and mid level roles on select teams in the United States, starting in the second half of 2026. The report says Google plans to extend the format to other teams if the pilot works. If you are interviewing outside those roles, or outside the United States, assume the pilot does not apply to you unless your recruiter says it does.
What about Google's other rounds?
The rest of the loop moved the other way. interviewing.io reported, in an edit to its survey dated 8 October 2025, that Google had finished a pilot of in person software engineer interviews at major engineering sites: two virtual interviews, then three or four in person, with a plan to roll this out for all software engineer roles within two to three months. Computerworld named Google among the companies bringing interviews back into the room to counter AI assisted cheating.
The algorithm interview is not going away soon. In the same survey none of the 52 interviewers at large tech companies said their company had dropped algorithm questions, though more than half said they had changed the kind of questions they ask.
Would using AI outside the pilot be cheating?
Yes, unless you are told otherwise. Nothing in the sources we checked allows AI in Google's other rounds, and the move to in person interviews was made precisely to stop hidden help. If a recruiter invites you to the pilot round, the invitation is your permission, and it covers that round only.
How should you prepare if you are in the pilot?
Practise the three verbs in the report: read, debug and optimise code you did not write. Start by mapping the code before asking the assistant anything, so your first question is specific. Validate every claim the assistant makes against a test or a quick experiment, and say out loud what you checked, because output validation is part of what is graded.
Our invoice totals round is built for exactly this: a small, realistic package with a bug report, failing tests and a trap that only reading the surrounding rules will catch. Run it with whatever assistant you have, under a 60 minute timer.
Then practise the unaided rounds too. A pilot round is one interview in a loop, and the classic algorithm and design rounds still decide most of the outcome.
What should you ask your recruiter?
Ask whether your loop includes the code comprehension round, which assistant and interface you will use, whether the assistant can edit files or only chat, and how long the round lasts. Ask whether any other round allows AI. The answers are worth more than any guide, including this one, because the pilot is new and its details may change.