AI in Tech Interviews: What to Allow, and How to Tell if a Candidate Is Using It

What the latest figures show

The Institute of Student Employers (ISE) published its Student Recruitment Survey on 14 October 2026. Of the 155 employers who took part, 92% had suspected or confirmed misconduct by graduate and apprentice applicants in the past year, up from 85% in 2025 and 78% in 2024. Undeclared use of AI in interviews was the most common form, reported by 66%, ahead of online assessments at 47%. Concerns about CVs and covering letters were far lower, at 10%.

Employers are responding by setting rules rather than banning AI. 82% now publish guidance on acceptable AI use, up from 55% a year ago, and 64% accept AI somewhere in the process, mostly for CVs and covering letters. Most draw the line at online assessments, interviews and assessment centres, and 77% are reviewing or redesigning how they assess candidates.

Why it matters for tech hiring

The ISE survey covers early-careers hiring, but the same tools are available to every candidate. In a technical interview, an AI assistant can suggest an answer to a coding question or a system design prompt in seconds, and in a remote video interview it can be hard to see. Our article on CV fraud in tech recruitment covers the more serious end of this, where the person interviewed is not the person who starts.

The risk for employers is simple: you hire someone on the strength of answers they could not give on their own, and you find out in the first few weeks of the job. From January 2027, unfair dismissal protection starts after six months' service, so getting the decision right at interview matters more. Our Employment Rights Act article covers what that means for probation.

Decide what you allow, and say so

Most tech professionals use AI tools at work, and many employers want them to. A blanket ban is hard to enforce and can put off good candidates. A clearer approach:

• Tell candidates in the interview invitation which tools they can use at each stage. For example, AI is fine for preparing a CV, but not during a live interview.
• If you allow AI in the job, you can allow it in a take-home exercise. Then ask the candidate to talk you through what the tool produced and what they changed.
• At the start of a remote interview, ask candidates to confirm that no assistant is running, and to share their screen for any coding task.
• Make reasonable adjustments where a candidate needs assistive technology. The Equality Act still applies.

Signs a candidate may be using AI live

None of these proves anything on its own, and nervous candidates do some of them too. Look for patterns:

• a pause, then a fluent, textbook answer that does not match how the candidate speaks elsewhere
• eyes moving to read from another part of the screen, or typing when no task has been set
• strong answers to standard questions, but difficulty with a follow-up about their own work
• a polished explanation of a concept with no real example of using it
• a sharp change in style between one answer and the next

If you have a concern, ask a follow-up rather than accusing anyone. "How would you debug that if it failed in production?" or "What did you get wrong the first time you built it?" quickly shows whether the knowledge is theirs.

Interview formats that are harder to fake

• A talk-through of the candidate's own work: a project, a pull request or an incident they handled. Ask why decisions were made and what they would change.
• Live problem-solving on a shared screen, where how they reason matters more than the final answer.
• Code review: a short snippet with a bug, and the question "what is wrong here, and why?"
• Follow-up questions that go two or three levels deep on anything they claim.
• An in-person final stage, or at least a face-to-face meeting before an offer for a remote role.

The AI and junior tech hiring article has more on assessing early-careers candidates now that AI can write the answer.

How we can help

Every candidate we put forward has had an in-depth conversation with a Deerfoot consultant and has been reviewed by two consultants before a client sees them. Once an offer is made, we verify identity and documents with certified TrustID checks. If you would like help designing a fair technical interview, send us the role.

Sources: Institute of Student Employers, Student Recruitment Survey 2026, reported by HRreview, 14 October 2026.

This article is a summary for employers, not legal advice.