Do candidates drop out of AI interviews?

More common than most hiring managers realise. Greenhouse surveyed 2,950 active job seekers across five countries. In the US, 63 per cent had been interviewed or assessed by AI in the previous twelve months, up 13 percentage points in six months. In Germany the figure was 57 per cent, in Australia 54 and in the UK 47.

Employers are adopting AI quickly across Europe too. According to Eurostat, 20.0 per cent of EU enterprises with ten or more people used at least one AI technology in 2025. That figure covers every use, not hiring alone, but it does mean a growing share of candidates will meet a system before they meet a recruiter.

Not because they object to technology. The report points to three things that are entirely within an employer's control.

Nobody told them. Of the US candidates, 70 per cent were never clearly told upfront that AI would be evaluating them, and one in five only found out once the interview had started. That feels like a trick, and a candidate who feels misled in round one has little reason to stay for round two.

There is no human in the room. The reason US candidates cite most often for abandoning a process is a pre-recorded video interview scored by AI with no person present. Someone who already has a job is not going to talk to a camera without knowing whether anyone will ever watch it.

They doubt it is fair. Across all five markets, between 63 and 70 per cent of candidates do not believe most employers use AI responsibly. That distrust is there before they reach your process, so you start the conversation behind.

And the people who leave first are the ones you most want to keep. Strong candidates have options and little patience for a process that treats them as input. The wider picture of why candidates drop out is covered under candidates drop out.

Start with what AI is genuinely good at: scheduling, summarising, preparing interviews and getting back to every applicant quickly. Our view on that is set out on the AI in recruitment page. Things go wrong where AI replaces the first human contact.

Say so in advance. Put in the invitation that AI is used, for what exactly, and who reviews the outcome. In Europe this is not only good manners; there are rules, summarised in which EU rules apply to AI in hiring now.

Keep a person in charge, and make that visible. Candidates accept an automated step far more readily when they know a recruiter reads the result and that they can speak to that recruiter.

Measure where people leave. Track how many candidates move on at each step, and look beyond the number of applications to who is still there at the end. If you already do that for your campaigns, as described under data and optimisation, you can do it for selection as well. What the steps should look like is covered in what does a good hiring process look like.

Are AI interviews allowed in Europe?

Yes, within limits. The GDPR already applies, emotion recognition in the workplace is banned, and stricter requirements are coming for systems that select or assess candidates. You can use an AI interview, but you need to be able to explain what it does, and a candidate should in principle not be rejected solely by a system without a person looking at it.

Will candidates simply get used to it?

Perhaps, but the numbers do not point that way yet. Exposure among US job seekers rose by 13 percentage points in six months, yet only 21 per cent of them believe most employers use AI responsibly and transparently. Familiarity does not fix poor execution: a candidate who never hears what happened to their recording is not going to feel better about it the next time.

Our team is stretched. Should we not automate more?

Automate the work around the candidate, not the contact with the candidate. Scheduling, feedback and note taking eat time, and AI really helps there. Handing over the first conversation saves time you later lose on candidates who dropped out. Other ways to relieve the pressure are under overloaded recruitment team.

Does this apply outside the five countries surveyed?

We do not know, and we will not pretend to. The study covered the US, the UK, Germany, Australia and Ireland. Distrust sits in the same range in all five, which is reason enough to ask your own candidates rather than assume your market is different.

At SocialFind we help you see where in your selection candidates drop out and why. If you would like articles like this in your inbox, subscribe to our newsletter.