{"id":348,"date":"2026-08-11T21:40:36","date_gmt":"2026-08-11T21:40:36","guid":{"rendered":"https:\/\/talently.tech\/en\/blog\/ai-breaking-technical-interviews-fake-candidates\/"},"modified":"2026-08-21T01:22:32","modified_gmt":"2026-08-21T01:22:32","slug":"ai-breaking-technical-interviews-fake-candidates","status":"publish","type":"post","link":"https:\/\/talently.tech\/en\/blog\/ai-breaking-technical-interviews-fake-candidates\/","title":{"rendered":"AI is breaking technical interviews: spot fake candidates in 2026"},"content":{"rendered":"Reading Time: <\/span> 6<\/span> minutes<\/span><\/span>\n

A candidate who can’t explain their own code isn’t new. What’s new is that in 2026 someone can clear a 45-minute algorithm screen without understanding a single line they typed, using nothing more exotic than a second monitor and a browser tab. If your interview loop was designed before 2023, it now measures the quality of a candidate’s tooling, not the quality of their engineering.<\/p>\n\n\n\n

\n

TL;DR<\/h2>\n
  • The problem isn’t AI use, it’s unverifiable authorship.<\/strong> You need to know who did the thinking, not which tools were open.<\/li>
  • Five attack patterns dominate: LLM-coached live coding, real-time answer overlays, proxy interviews, AI-polished r\u00e9sum\u00e9s, and deepfaked video candidates.<\/strong> Each has different detection signals.<\/li>
  • The strongest single filter is a deep “why” follow-up<\/strong>: change a requirement mid-exercise and watch whether the candidate reasons or re-prompts.<\/li>
  • Redesign beats surveillance. Live debugging of unfamiliar code, pair-programming in your real repo, and a take-home plus defense<\/strong> are nearly immune to overlay tools.<\/li>
  • Verify identity once, early, and cheaply: government ID matched against the live face, plus a recorded 60-second intro<\/strong> you can compare across every later call.<\/li>
  • Do not ban AI and do not install invasive proctoring. Both cost you strong candidates<\/strong> and neither stops a competent cheater.<\/li><\/ul>\n<\/div>\n\n\n

    The five things actually happening in your loop<\/h2>\n\n\n

    LLM-coached live coding<\/strong> is the most common. The candidate shares their screen, an assistant on a second device or a stealth overlay reads the problem, and the answer arrives in 10-20 seconds. Commercial “undetectable” interview copilots have existed since 2024 and now run as always-on-top windows invisible to screen capture. Screen sharing proves nothing.<\/p>\n\n\n\n

    Proxy and impersonation interviews<\/strong> are worse, and rarer, but they’re the ones that get someone hired who cannot do the job. A strong engineer sits the technical rounds; a different person shows up on day one, or the same person is being fed everything by a colleague off-camera. In its 2024-2025 advisories the US DOJ documented hundreds of American companies unknowingly employing North Korean IT workers using stolen or borrowed identities. This is not a theoretical category.<\/p>\n\n\n\n

    Deepfaked video<\/strong> has crossed the threshold where a live face swap runs on consumer hardware at 30fps. AI-polished r\u00e9sum\u00e9s<\/strong> are the highest-volume, lowest-severity version: a real person with thinner experience, whose CV has been rewritten to match your job description keyword for keyword. All five are one failure: your process rewards output you cannot trace back to a mind.<\/p>\n\n\n

    Detection signals that hold up<\/h2>\n\n\n

    Most “tells” you’ll read about are noise. These are the ones that survive contact with real interviews:<\/p>\n\n\n\n

    Signal<\/th>What it looks like<\/th>What to do next<\/th><\/tr><\/thead>
    Latency inversion<\/strong><\/td>Long pause, then a fluent, complete, well-commented answer delivered at once<\/td>Interrupt mid-answer with a constraint change<\/td><\/tr>
    Eye-line drift<\/strong><\/td>Consistent glance off-camera before each answer, on a fixed rhythm<\/td>Ask a question requiring them to draw or point at a shared screen<\/td><\/tr>
    Register mismatch<\/strong><\/td>Polished textbook vocabulary for concepts, vague language for their own project<\/td>Ask about a decision they regret in that project<\/td><\/tr>
    Zero debugging instinct<\/strong><\/td>Code is correct on the first try, but they cannot find a bug you inject into it<\/td>Break their solution and ask them to diagnose it<\/td><\/tr>
    Audio-visual lag<\/strong><\/td>Lip sync drift, hair\/glasses edge artifacts, unnatural lighting stability<\/td>Ask them to turn their head 90 degrees or pass a hand across their face<\/td><\/tr>
    R\u00e9sum\u00e9 density<\/strong><\/td>Six roles, all with quantified outcomes, all phrased identically<\/td>Ask for the raw numbers behind one metric and how it was measured<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

    One rule matters more than the list: an assisted candidate is optimized for producing answers, not for handling change.<\/strong> Any perturbation (new constraint, wrong assumption, sudden requirement) breaks the loop between the candidate and their helper. Introduce one perturbation per exercise and score how fast they recover.<\/p>\n\n\n\n

    Also stop treating speed as a proxy for skill. In an AI-assisted world, the candidate who solves your LeetCode-style problem in four minutes flat is more<\/em> suspicious than the one who spends twenty minutes talking through tradeoffs.<\/p>\n\n\n

    Redesign the loop, don’t police it<\/h2>\n\n\n

    The fastest fix is not a detection tool. It’s an exercise that an LLM cannot complete without the context that lives in your team’s head.<\/p>\n\n\n\n

      \n
    • Live debugging of unfamiliar code.<\/strong> Give them 150-300 lines from a real service with two seeded bugs: one obvious, one a race condition or an off-by-one in a boundary case. Assistants are good at writing code and mediocre at localizing failures inside unfamiliar systems. Budget 30 minutes.<\/li>\n\n\n\n
    • Pair-programming in your actual repo.<\/strong> A small, real ticket, your conventions, your test suite, your CI. Let them use Copilot or Claude openly. You are scoring how they direct the tool, reject bad suggestions, and navigate a codebase they’ve never seen.<\/li>\n\n\n\n
    • Take-home plus defense.<\/strong> The take-home is cheap to fake and that’s fine. It’s a filter, not a signal. The 30-minute defense is the signal: “Why this data structure?”, “What happens at 100x load?”, “Delete this abstraction and tell me what breaks.”<\/li>\n\n\n\n
    • Architecture with a moving target.<\/strong> Fifteen minutes into a design discussion, change a core assumption. Reads go from 1k\/s to 200k\/s. The write path must now be strongly consistent. Real seniors visibly re-plan; coached candidates restate their original answer with new adjectives.<\/li>\n<\/ul>\n\n\n\n

      Score authorship explicitly. Add a line to your scorecard: “Could this person rebuild what they showed me, alone, tomorrow?”<\/strong> Make it a required field. It changes what interviewers pay attention to more than any tooling change will.<\/p>\n\n\n

      Verify identity once, properly<\/h2>\n\n\n

      Identity checks belong at the start of the funnel, not at the offer stage where the switch has already happened.<\/p>\n\n\n\n

      Match a government photo ID against the live face<\/strong> on the first video call, and note the name on the ID. Record a 60-second intro clip<\/strong> in that first call and keep it. Every later interviewer compares against it, which is what catches the classic bait-and-switch between screen and final round. Confirm that the video call name, email domain, and LinkedIn history<\/strong> are consistent, and that the profile has connections and history predating your job posting by years.<\/p>\n\n\n\n

      For final rounds, require camera on and one unbroken session<\/strong>, and ask a question that forces natural head movement. If you work with a recruiting partner, ask them directly what identity verification they run, whether they’ve ever caught a substitution, and what happened. A partner who has never seen the problem either isn’t looking or isn’t screening at volume.<\/p>\n\n\n

      What not to do<\/h2>\n\n\n

      Don’t ban AI.<\/strong> You’ll filter out the engineers who use it well (which is now most of the strong ones) and keep the ones who memorized answers. State a policy instead: AI allowed in take-homes and pairing, disclosed; not allowed in the 30-minute defense, where you’re testing understanding.<\/p>\n\n\n\n

      Don’t install invasive proctoring.<\/strong> Keystroke logging, eye-tracking, forced room scans, and remote-desktop monitoring have three problems: senior candidates drop out rather than submit to them, they generate false positives against anyone with ADHD or a small apartment, and they don’t catch the second device, which is where the cheating actually happens. You get a worse pipeline and no additional integrity.<\/p>\n\n\n\n

      Don’t over-index on AI-detection scores.<\/strong> Text classifiers are unreliable on short samples and biased against non-native English writers, which for LATAM, Eastern European, and Asian candidates means systematically penalizing good engineers. Never reject on a detector score alone.<\/p>\n\n\n\n

      Don’t add rounds.<\/strong> Longer loops lose the candidates with options and give a determined faker more practice. Four well-designed hours beat eight defensive ones.<\/p>\n\n\n

      Frequently Asked Questions<\/h2>\n\n

      Isn’t it fine if a candidate uses AI to solve the interview problem (they’ll use it on the job too)?<\/h3>\n\n\n

      Yes, and that’s exactly why the exercise has to change. On the job they’ll use AI inside your codebase, under review, with consequences. Test that: open tools, real repo, real ticket, and score judgment (what they accept, reject, and verify), not whether the final code compiles.<\/p>\n\n\n

      What’s the single highest-leverage change if I can only make one?<\/h3>\n\n\n

      Add a 30-minute defense round to whatever you already do. No new code, just deep “why” questions about work the candidate already submitted, plus one mid-conversation requirement change. It costs half an hour per finalist and catches the large majority of coached candidates.<\/p>\n\n\n

      How do I spot a deepfake on a normal video call?<\/h3>\n\n\n

      Ask for movement the model struggles with: turn the head fully in profile, pass a hand slowly in front of the face, hold a hand next to the cheek. Watch for edge artifacts around hair and glasses, lip-sync drift under fast speech, and lighting that doesn’t change when they move. Real-time face swaps degrade under occlusion and extreme angles.<\/p>\n\n\n

      Will these changes make our process feel hostile to good candidates?<\/h3>\n\n\n

      Only if you frame it as suspicion. Frame it as realism: “You’ll pair with us in our repo on a real ticket, tools allowed.” Strong engineers generally prefer this to whiteboard trivia, and it doubles as a preview of the work, which helps close them later.<\/p>\n\n\n

      How much does interview fraud actually cost when it gets through?<\/h3>\n\n\n

      Salary is the small part. The real cost is time: these hires surface in month two or three, after they’ve burned a senior engineer’s review cycles, and then you restart a search that takes 30-60 days. On a ten-person team that’s roughly a quarter of lost roadmap.<\/p>\n\n\n

      Should we still run take-home assignments?<\/h3>\n\n\n

      Yes, but demote them. Treat the take-home as a cheap filter and never as evidence of skill, keep it under three hours, and make the defense mandatory. If you’re not willing to run the defense, drop the take-home entirely: an undefended take-home in 2026 tells you nothing.<\/p>\n","protected":false},"excerpt":{"rendered":"

      Reading Time: <\/span> 6<\/span> minutes<\/span><\/span>In 2026 a candidate can clear a 45-minute algorithm screen without understanding a line of it. The five fraud patterns, the detection signals that hold up, and the interview redesign that beats surveillance.<\/p>\n","protected":false},"author":3,"featured_media":212,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"inline_featured_image":false,"footnotes":""},"categories":[6],"tags":[],"class_list":["post-348","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-hiring-challenges"],"_links":{"self":[{"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/posts\/348","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/comments?post=348"}],"version-history":[{"count":6,"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/posts\/348\/revisions"}],"predecessor-version":[{"id":402,"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/posts\/348\/revisions\/402"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/media\/212"}],"wp:attachment":[{"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/media?parent=348"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/categories?post=348"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/talently.tech\/en\/blog\/wp-json\/wp\/v2\/tags?post=348"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}