Delhi Police have arrested seven people, including the alleged mastermind, in connection with a sophisticated interstate racket that used hidden cameras, nano earbuds, and ChatGPT to help candidates cheat in government recruitment examinations. The alleged kingpin, Sonu Kumar, was a firefighter deployed with the Chandigarh Fire Department, and the syndicate reportedly charged aspirants up to ₹12 lakh, supplying specially modified electronic gadgets that let candidates inside exam halls secretly receive answers transmitted from outside.
The case surfaced after Central Sanskrit University lodged a complaint on July 26 alleging unfair means during a Lower Division Clerk (LDC) recruitment exam held at a Kendriya Vidyalaya in Delhi's Janakpuri. During frisking, officials recovered a customised mobile phone from one candidate, while another was found carrying a Bluetooth device fitted with a SIM card along with concealed nano earbuds; 213 candidates had appeared for just six vacancies.
Investigators found that miniature cameras were fitted near shirt buttons, allowing candidates to discreetly photograph question papers, which were then transmitted to an accomplice outside the centre. That accomplice allegedly fed the images into ChatGPT to generate answers, which were then relayed back to candidates through the nano earbuds in real time. Police have so far identified 10–12 candidates who allegedly used the racket's services, though the actual number could be higher. Further surveillance led police to Charkhi Dadri, Haryana, where another accused, Sonu Maan, was apprehended while allegedly facilitating cheating in a separate government exam, with a tablet linked to a foreign WhatsApp account and additional nano earbuds recovered from his possession.
This case fits a pattern that has now surfaced repeatedly across India in 2026, not an isolated incident. A nearly identical modus operandi — photographing question papers, feeding them into ChatGPT, and relaying AI-generated answers back via hidden earpieces — was used in the Telangana State Public Service Commission paper-leak case, where a state power company engineer allegedly helped multiple candidates cheat in Assistant Executive Engineer and Divisional Accounts Officer exams using Bluetooth micro-earpieces, and again in Maharashtra's Class 12 board exams, where teachers and exam staff were booked for uploading photographed question papers to ChatGPT and distributing printed AI-generated answers to students.
What makes this trend significant isn't the novelty of exam cheating — question paper leaks and hidden devices predate generative AI by decades — but the speed and scale AI adds to an old crime. Previously, an accomplice outside an exam hall needed genuine subject-matter expertise to solve questions correctly under time pressure. Now, that expertise requirement is effectively removed: anyone with a smartphone and ChatGPT access can generate competent answers to a wide range of exam questions in seconds, regardless of their own knowledge of the subject. This dramatically lowers the skill barrier to running a cheating syndicate, which likely explains why these rackets have scaled to charge candidates lakhs of rupees and operate across state lines.
It also exposes a detection gap examination bodies haven't fully closed. Physical security measures (frisking, hidden-camera detection, jammers) were designed around older cheating methods — chits, whispered answers, physical documents — not the combination of miniaturized transmission hardware and instant AI-generated answers. The recurring involvement of insiders (a fireman, a power-company engineer, teachers and exam-hall peons) also points to a structural weakness: technology alone doesn't explain how devices repeatedly get past frisking and question papers get photographed within minutes of being unsealed — that requires someone with institutional access looking the other way, or actively participating.
For recruitment bodies and examination authorities, this is likely to accelerate the same trend we've discussed elsewhere: a push toward stronger identity and behavioral verification at the point of examination — device-detection sweeps, signal jamming, and potentially biometric/liveness monitoring during exams — treating high-stakes recruitment tests with the same threat model as financial fraud, rather than relying on physical frisking alone.
