Leave a Review & Get 30% OFF - Limited Time Offer!

00:00:00
Guides

AI frauda statistici interviu 2026

Last updated: January 2, 2026|6 min read|By InterviewMan Team

AI Interview Cheating Statistics 2026: The Complete Data

ok so i have been collecting every stat i can find on AI interview cheating for the past month because i kept seeing the same numbers cited without sources and a fost driving me nuts. Half the articles out there reference "studies show" without linking a single study. So here is every real number i could verify with an actual source, organized by what it tells you about where this is heading. Fair warning this is long but i wanted everything in one place because i nu am putut find that anywhere else.

The headline number that keeps getting passed around is from Fabric's analysis of over 50,000 candidates between June and December 2025. Cheating adoption more than doubled from 15% to 35% in that six month window. Let that sink in for a second, asta e not a gradual trend, asta e a doubling in half a year. Their more detailed study of 19,368 interviews found that 38.5% of all candidates were flagged for cheating behavior, with technical roles hitting 48% and sales roles at 12%, roughly a 4x difference. Junior candidates with zero to five years of experience cheated at nearly double the rate of senior professionals.

The scariest part of the Fabric data is what happens when cheaters are not caught. 61% of cheaters scored above passing thresholds of 7.0 or higher and would have advanced to the next round without detection. That means more than half the people using AI assistance are getting through, and companies only catch them later when performance nu match the interview or they nu catch them at all.

Repeat behavior patterns are wild too. Among candidates who interviewed multiple times, 47% never cheated across any interview, 30% cheated in every single one, and 23% were situational cheaters who sometimes did and sometimes nu am. The 30% who cheat every time are practic treating it as standard operating procedure at this point.

For how they are doing it, Fabric broke it down by method. Dedicated AI assistants like Cluely and Interview Coder accounted for 45% of cheating. LLM voice mode, practic talking to ChatGPT on a al doilea dispozitiv, was 34%. Tab switching and secondary screens was 18%. Live human help was only 3%. The fact that people are moving to dedicated tools in loc de just googling answers on a second tab tells you this is becoming professionalized.

Karat observed a five-fold increase in cheating detection rates over the past two years which tracks with the Fabric numbers. A separate survey found that 59% of hiring managers now suspect candidates of using AI tools to misrepresent their abilities during live assessments, and 62% of hiring professionals admitted that candidati in cautarea unui job are now better at faking with AI than recruiters are at detecting it. That 62% number is the one that should worry companies the most.

On the candidate side the numbers are even more striking. A survey found that 71% of recent candidati in cautarea unui job admitted to cheating during the procesul de angajare with methods ranging from Googling answers to using AI generators. One tech leader reported that 80% of candidates used an LLM to complete their top-of-funnel code test even though they were explicitly told not to.

The temporal patterns in the data are interesting. Fabric found a 3x increase in cheating from July to September 2025, with a sudden spike in late 2025 marking what they called a shift from experimental to structural cheating. Sunday interviews showed the cel mai mare cheating rate at 47.1% with other weekdays clustering between 35 and 40%. The Sunday number are sens, people interviewing on weekends are probably doing it from home with nobody watching.

The financial impact of a bad hire is well documented. Direct costs for a single poor hire in a hundred fifty thousand dollar engineering role range from 30 to 150% of first year earnings, so fifty thousand dollars or more. Average time to fill for technical roles before a restart is 42 days. Companies that hire someone who cheated their way in are looking at months of underperformance, a restart of the procesul de angajare, and potentially six figures in wasted salary and lost productivity.

Looking forward, Gartner projects that by 2028 entirely fabricated candidate profiles will make up 25% of the candidate pool. One in four profiles being fake is a number that would just destroy remote hiring completely if it comes true.

The deepfake side of this is accelerating too. 17% of HR managers in a mid-2025 survey said they had directly encountered deepfake technology in a video interview. Deepfake incidents in Q1 2025 alone hit 179 cases, surpassing the entire 2024 total. AI-generated deepfake scams rose 700% in 2025, and deepfake-related fraud losses in the US tripled from 360 million to 1.1 billion dollars.

For context on the tools people are using, exista roughly twenty plus interview AI tools on the market chiar acum ranging from extensie de browsers to aplicatie desktops. Pricing goes from twelve dollars a month for InterviewMan on the planul anual up to two hundred ninety nine a month for Interview Coder which only covers runda de codings. The Cluely bresa de date in mid 2025 exposed 83.000 de utilizatori including which interviews each person used the tool in, after hackers found an admin password in a public GitHub repo. That breach really showed that not all tools de fapt protect their users.

i use InterviewMan myself and the distinction asta conteaza to me is between tools that protect your data and tools that leave you exposed. InterviewMan runs as a aplicatie desktop, not a extensie de browser, so nu apar in proctoring logs. 57.000 de utilizatori, 4.8 stele din 257 de evaluari, twenty plus functii stealth, twelve bucks a month annual. Across all those users i nu am putut find a single detectie confirmata report on any proctored platform. Whether you think using these tools is ethical or not the data makes one thing clear, the percentage of candidates using them is only going up and pretending otherwise just puts you at a disadvantage against the 35% who are not pretending.

Ready to Ace Your Next Interview?

Join 57,000+ professionals using InterviewMan to get real-time AI assistance during their interviews.

ShareTwitterLinkedIn

Related Articles

Try InterviewMan Free

AI interview assistant. Undetectable.

Get Started