TL;DR
Analysis of 19,368 AI interviews reveals that cheating has become a structural problem in hiring, with rates jumping 3x in late 2025.
- 38.5% of all candidates flagged for cheating behavior
- Technical roles show 48% cheating rates vs 12% for sales
- 61% of cheaters score above pass thresholds and would advance without detection
- 3x increase in cheating from July to September 2025
- Junior candidates (0-5 YoE) cheat at nearly double the rate of senior candidates
Introduction
AI tools have fundamentally changed how candidates take interviews. Real-time assistance during live interviews, once rare and risky, is now common and nearly invisible. Tools like Cluely, Interview Coder, and Final Round AI use invisible screen overlays and audio transcription to feed candidates answers without appearing on screen shares.
This report analyzes 19,368 AI-powered interviews conducted on Fabric's platform between July 2025 and January 2026. Every interview included automated cheating detection through behavioral and technical signals.
The findings show that cheating is no longer an edge case. It is concentrated in specific roles, effective enough to beat traditional interview scoring, and growing rapidly.
How Big is the Interview Cheating Problem?
Roughly 4 in 10 candidates are cheating in interviews
Interview cheating has moved from isolated incidents to a systemic issue. Across nearly 20,000 interviews, more than a third of candidates showed cheating behavior.
- Normalization of AI - Just as spellcheck became standard for writing, candidates now view AI assistance as a standard tool for interviewing. The moral barrier has lowered significantly as generative AI has become a daily utility in their actual work.
- The Prisoner's Dilemma - Many genuine candidates feel forced to cheat because they assume their competition is already doing it. They fear that being the only honest participant puts them at an unfair disadvantage in a market where efficiency and speed are rewarded.
Cheating rates jumped 3x in late 2025
The sudden spike in late 2025 signals a shift from experimental to structural cheating. This was likely driven by two factors:
- The Viral Effect - Social media platforms saw an explosion of content showing how to beat interviews using AI. Once candidates saw others getting offers with these tools, the Fear Of Missing Out (FOMO) drove mass adoption.
- Tool Maturity - Late 2025 saw the release of stable, invisible overlay tools. Before this, cheating was risky and buggy. The new generation of tools made it feel safe and easy, lowering the barrier to entry
What this means for your hiring process
Cheating is now a default assumption for high-volume hiring.
- Start tracking AI interview cheating - If you are not actively detecting, assume at least 1 in 3 candidates is getting external help
- Rethink your interviews - Interview processes designed before the AI era may not account for current candidate behavior
Which Candidates Are Most Likely to Cheat?
Cheating rates vary dramatically by role type, experience level, and compensation.
Technical roles cheat at 4x the rate of sales roles
Technical and operations roles most commonly see AI based interview cheating. This can be because of two reasons -
- Better tech savviness - Candidates in these roles are more tech savvy and are more likely to try a new technology
- Better cheating results - Interviews in these roles have more objective questions which are easier to cheat on. Interviews for roles like sales, marketing and HR include more open ended questions with a higher focus on soft skills.
All technical subroles have high cheating rates
No technical specialization is immune. The elevated cheating signal is broad-based across all engineering functions.
Junior candidates cheat twice as much as senior candidates
All roles:
Candidates with 0-5 years of experience are significantly more likely to use assistance.
- High-Stakes Competition - The entry-level market is incredibly saturated. Junior candidates often feel that they need every possible advantage just to get a foot in the door.
- Knowledge Gaps - Unlike senior engineers who might use AI to speed up syntax they already know, junior candidates often use AI to generate answers for concepts they have never learned. With companies insisting on unrealistic knowledge expectations from junior candidates, interview cheating clears their path to getting a job.
There is negative correlation between cheating and comp
There is a clear downward trend where candidates in lower salary bands cheat at higher rates.
- Junior Roles - Most roles in this category fall under junior roles, and we previously discussed why cheating is more prevalent in younger professionals.
- Volume Approach - Candidates in this bracket often apply to hundreds of roles. They use AI tools to mass-interview efficiently, treating it as a numbers game rather than a curated career move.
What this means for your hiring process
Prioritize detection based on your hiring mix. A company hiring mostly engineers has 4x the cheating exposure of one hiring salespeople. If you are scaling junior technical hiring, detection is essential.
- Stop using standard LeetCode-style questions - These are the easiest for AI to solve; switch to vague, real-world debugging scenarios or system design discussions where there is no single "correct" code block.
- Assume secure browsers are compromised - Do not rely on browser locking or tab-tracking software; assume candidates can bypass these and focus on behavioral signals instead.
- Create an AI use policy for interviews - Align within your team on what type of AI usage is okay with and what is not. This will help you to better educate your team as well as your candidates.
The Impact of Interview Cheating on Hiring Process
Cheating works. Candidates with external help perform better on interview rubrics. Without a separate cheating signal, interview scores cannot distinguish genuine skill from assisted performance.
Cheaters score higher than non-cheaters
Cheating works. Most candidates who use these tools score high enough to be hired if no separate detection exists.
- Rubric Hacking - AI models are trained on the exact same documentation and textbooks that hiring managers use to create scoring rubrics. This means the AI generates the textbook-perfect answer that interviewers are conditioned to look for.
- False Confidence - Having the answer displayed on a screen eliminates the stuttering and nervousness typical of difficult questions. This artificial smoothness boosts their communication scores, helping them pass soft-skill checks.
Using interview score ≥ 7.0 as the pass threshold, 61.1% of flagged cheaters would advance through your hiring process. Interview performance alone does not filter them out.
What this means for your hiring process
Interview scores are necessary but not sufficient. You need a separate cheating signal. If you are seeing candidates who interview well but underperform on the job, undetected cheating is a likely explanation.
- Redefine your pass criteria - Stop grading solely on the final answer; start grading the journey. A perfect solution with zero backspaces or hesitation should be treated as suspicious, not excellent.
- Adopt cheating detection tools - Do not just trust on interview scores. Adopt solutions that help you prevent and detect cheating for better hiring outcomes.
Other Interesting Data from our Analysis
The data reveals patterns in how cheating happens.
30% of repeat candidates always cheat
For candidates who interviewed multiple times on Fabric, cheating behavior falls into three distinct patterns. 47% never cheat across any interview, while 30% cheat in every single interview they take. The remaining 23% are situational cheaters.
This split suggests two different cheating mindsets:
- Cheating as a fixed strategy - For the 30% who always cheat, this is a deliberate approach to interviews. They have invested in tools, learned the workflow, and apply it consistently regardless of role or company.
- Cheating as a situational response - The 23% who sometimes cheat likely respond to context: perceived difficulty of the role, time pressure, or how high-stakes the opportunity feels. They may cheat for dream jobs but interview honestly for backup options.
Cheating distribution across the week
Sunday has the highest cheating rate at 47.1%, while the other days cluster between ~35-40% with small statistical variance.
- Sunday is the worst day - Weekend interviews often happen from home with fewer distractions or observers. Candidates have more freedom to set up cheating tools, use secondary devices, or speak answers aloud without concern.
Most common ways candidates cheat
Dedicated cheating assistants like Cluely and Interview Coder account for 45% of cheating cases, followed by voice mode on LLMs like ChatGPT at 34%. Traditional methods like tab switching or using a second screen make up 18%, while live help from another person is rare at just 3%.
- Invisible tools dominate - The top two methods (79% combined) are designed to be undetectable by screen sharing. These tools use invisible overlays or audio-only input, making them impossible to catch through traditional proctoring.
- Old methods are dying out - Tab switching and second screens are now minority tactics. Candidates have learned these are easily flagged, so they have migrated to purpose-built cheating software.
- Human accomplices are impractical - Live help from a friend requires coordination and introduces lag. Automated tools are faster, more reliable, and do not require scheduling another person.
How to Detect Interview Cheating
If cheating has become so commonplace, then the natural question that arises is how to detect and prevent it.
Traditional proctoring flags tab switches or second faces. This approach is easily bypassed by modern tools using invisible overlays. Effective detection requires analyzing behavioral signals throughout the interview.
Watch for the Lag Loop
AI tools create a consistent 3-5 second delay after every question: audio capture, AI processing, candidate reading. In normal conversation, response time varies with question difficulty. In cheated interviews, response time stays identical regardless of complexity.
Look for reading eye movements
Thinking eyes drift upward or to the side. Reading eyes move in straight horizontal lines left to right, then snap back. If a candidate's eyes follow mechanical reading patterns while supposedly thinking, they are likely reading from an invisible overlay.
Design interviews that break cheating tools
Cheating tools thrive on standardized questions. When a candidate provides a polished answer, drill down: ask for a specific failure example. Ask about technologies that do not exist. AI tools will hallucinate answers; genuine candidates will admit unfamiliarity.
Use AI-powered platforms with built-in detection
Platforms like Fabric analyze 20+ signals during live interviews: gaze tracking, response timing, keystroke dynamics, language patterns. Based on extensive evaluation, Fabric detects cheating in 85% of cases with timestamped evidence.
Conclusion
Cheating is widespread (38.5%), concentrated in technical and junior hiring, and effective enough that most cheaters pass traditional interviews. The 3x jump in late 2025 shows this is now the new normal.
Detection succeeds by making interviews more adaptive and analyzing behavioral signals that cheating tools cannot mask. For companies hiring at volume, the question is no longer whether candidates can do the job. It is whether the candidate is real.
Methodology
Dataset: 19,368 interviews, July 2025 - January 2026 | Platform: Fabric AI interview platform | Cheating threshold: Probability > 40% | Pass threshold: Interview score ≥ 7.0 | Salary currency: INR CTC
FAQ
How does Fabric detect interview cheating?
Fabric analyzes 20+ signals during live interviews including gaze patterns, response timing, and language analysis. These combine into a cheating probability score with timestamped evidence. Read more about it here.
What is Fabric?
Fabric is an AI-powered interview platform that conducts automated interviews with live coding, case studies, and role plays, with built-in cheating detection.
Can cheating tools really be invisible to screen sharing?
Yes. Modern tools use low-level graphics hooks to render overlays that exist only on the local display, invisible to screen share video streams.
What should I do if I suspect a candidate cheated?
Review behavioral signals like response timing and eye movement patterns. Ask unexpected follow-up questions requiring genuine experience to answer.
Are take-home assessments safer than live interviews?
No. Take-home assessments have higher cheating rates because candidates have unlimited time and privacy. Live interviews with real-time detection provide more reliable signals.