How AI Became a Cheating Tool for Tech Interviews

He is Roy Lee, a student kicked out of Columbia University for helping software developers cheat during their job interviews. But this student is not the only one doing this. Some people are using AI to gain what could be considered an unfair advantage. To answer the question, is using AI for job interviews cheating? The use of AI to cheat in job interviews is on the rise. It is estimated that nearly 50% of candidates for software engineering positions And educators across the country are raising the alarm over a new technology that is making it easier to cheat. Companies like Google, which often conduct virtual job interviews, are considering returning to in-person interviews to prevent AI-based cheating. So how did AI become a cheating tool for tech interviews? And how is this affecting the tech job market? In 2019, the pandemic forced companies to move their hiring processes to the virtual space. What initially seemed like a temporary solution, quickly became a common practice due to its multiple advantages like cost reduction, greater agility, and broader access to global talent. However, this format also facilitated a new type of cheating, the use of artificial intelligence tools out of the interviewer's sight. A job applicant was caught lip-syncing while someone else answers questions in a Skype interview in India. With the arrival of advanced language models, candidates began relying on these tools to solve complex technical problems in real-time during interviews. These solutions, which can generate code, fix errors, and provide detailed explanations in seconds, have changed the rules of the game in technical assessments. What was once a test of pure skill is now a test of who can best use the available technology. LeetCode is an online platform widely used by programmers to practice algorithm and computational logic problems. It is designed to help them prepare for technical interviews at tech companies. It is estimated that more than 70% of applicants for technical jobs in Silicon Valley use it as a fundamental part of their training. However, in recent years, this preparation process has been radically altered by the rise of AI tools like Claude and Codex, which now not only help with studying, but are also actively used to cheat during technical interviews. So I think I found an AI cheat code. This is a combination of Perplexity, DeepSeek, and Claude all in one interface. This occurs especially in remote interviews, where candidates can easily access external assistance without being detected. A study showed that one in five processes displayed clear signs of AI-generated assistance, and it is estimated that 27% of respondents admitted to using AI to answer technical questions in real time. This practice has been systematized by platforms like InterviewCoder, founded by Roy Lee, a 21-year-old Columbia University student. He decided to turn his experience with technical interviews into a business that provided live answers to candidates through covert tools, charging between $200 and $1,000 per session. Roy has developed a tool called InterviewCoder that basically uses AI to help job applicants to big tech companies cheat on their interviews. InterviewCoder openly presents itself as a service that helps candidates cheat undetectably. With provocative statements on its website, Lee not only challenges traditional methods of technical evaluation, but also harshly criticizes the system that upholds them. Lee argues that if tech companies call themselves AI first, they should also accept the use of AI during interviews. According to him, practicing for hundreds of hours on platforms like LeetCode led him to burnout and almost made him quit programming. The platform assisted in more than 1,000 interviews in less than a year, with a hiring rate of 65%, including positions at Amazon and Microsoft. The popularity of InterviewCoder has put major tech companies on high alert. I'm sharing my entire screen, and I'm going to show you guys how to use InterviewCoder. Google, for example, has considered returning to in-person interviews to prevent AI-based cheating. Amazon and TikTok withdrew internship offers after discovering these kinds of practices. Other companies like Deloitte and Anthropic have started to incorporate explicit policies against the use of AI during the hiring process. However, these responses reveal a structural dilemma. On one hand, companies promote the use of AI to boost productivity. Google claims that more than 25% of its new code Coding by AI, that's not exactly a new idea, but the scale of how it's being used at Google, more than 25% of all new code, that is unexpected and could have implications for the rest of big tech. On the other hand, they condemn its use during entry stages. This contradiction has created a rift between candidates and recruiters, where mistrust has become the norm. Although InterviewCoder was shut down after an investigation, the void was quickly filled by at least a dozen similar services operating from Asia and Eastern Europe, taking advantage of the lack of global regulation. The problem is amplified by the effectiveness of models capable of solving over 92% of LeetCode's hard problems in just seconds. These tools generate solutions in multiple programming languages, often accompanied by detailed explanations, becoming extremely useful tools, but also dangerous when used with deceptive intent. The impact is not only technical, but also organizational and ethical. Companies now face a new paradox. Candidates who shine in technical interviews, but fail at their tasks once hired. A recruiter from Amazon revealed that he hired an engineer who solved all the interview challenges, but didn't know how to use Git on their first day of work. This has led companies like Google to implement closed environments while Microsoft trains reverse AI models to detect artificially generated code. According to internal data leaked in 2024, 18% of codes submitted in technical assessments showed signs of being AI-generated. In addition to software-based cheating, there have also been documented cases of fraud involving invisible earpieces, smartwatches, and glasses with cameras. To the next questions, I would like you to share with me your screen immediately. Now. Uh, sure, sure. Uh, okay. Uh, can I use my screen? Yes. There's even identity substitution, known as proxy interviewing, where an expert takes the interview on behalf of the real candidate. According to CyberArk, in 2024, more than 200 documented incidents of technical interview fraud were recorded in North American companies alone. This situation poses not just a technological problem, but also an ethical and structural one. Many candidates justify the use of AI as a way to level a system they see as unfair or unrepresentative of the real skills required in day-to-day work. Meanwhile, the use of these technologies continues to spread unchecked. Companies are being forced to rethink their technical evaluation methods, questioning whether traditional algorithmic challenges truly measure what matters in professional development. Artificial intelligence is blurring the line between genuine talent and the ability to delegate to a machine. The challenge is not only to detect cheating. It is also to redefine what we understand as merit and competence in an industry that rewards speed, efficiency, and increasingly, the ability to collaborate with automated tools. As AI tools become a routine part of software engineering, their presence in candidate workflows is no longer exceptional, but expected. If evaluation criteria are not updated, the system runs the risk of rewarding those who cheat best instead of those who truly know how to build solutions.