How do you ensure candidates don’t use AI to cheat on technical interviews?

UPFLINX

This question was originally asked by one of our webinar attendees and has been shared here by the UPFLINX team to invite broader insights from the community.

ow do you make sure candidates aren’t using ChatGPT or other AI tools to pass technical assessments or coding interviews? It’s becoming harder to tell whether answers are genuinely theirs or just well-prompted responses. Curious how others are tackling this.

Julia

FlorianRocktaeschel

I studied in Aachen, and during remote exams they required three webcams, one on your face, one behind you, and a screen share. That level of monitoring isn’t practical (or recommended) for most interviews.

Instead, we suggest avoiding fully automated tests. Candidates can always Google answers or use AI tools. So what do you do? Ask follow-ups. Ask: Why did you choose this solution? Can you think of another way to solve it? AI can give answers, but it can't justify decisions like a human can, at least not consistently.

Focus on understanding how the candidate thinks, not just what they say. The goal is to assess reasoning, not regurgitation.

LauraVoinea

The way we ask questions now matters more than ever.

Borrow from teaching: ask layered questions. One correct answer isn’t enough, ask them to build on it, compare approaches, or explain trade-offs. If a candidate is relying too heavily on AI tools, they’ll quickly stumble when you move off-script.

Also, remember: some roles do involve using AI tools. If that’s the case, test how they use it. Can they write a good prompt? Can they critique a generated response? That’s a valid skill now.

Ultimately, even if someone passes with help, the probation period will reveal whether they’re truly capable. But by designing smart, structured questions, we can reduce that risk upfront.