Rebuilding trust the real challenge, says IIT Madras director in exam task force

Mumbai: The paper leak for India's undergraduate medical entrance examination, the 2026 National Eligibility-cum-Entrance Test (NEET), exposed a vulnerability few had anticipated: from the question paper setters, not the distribution process.

V. Kamakoti, director at Indian Institute of Technology (IIT) Madras and a member of the Centre's examination reforms task force, says the focus now is on tightening the vetting of paper setters, expanding use of artificial intelligence (AI) in surveillance, and examining whether NEET should move to multiple computer-based test sessions like the Joint Entrance Examination (JEE) does for engineering courses.

But beyond technology, rebuilding trust remains the toughest challenge, says the professor in an interview with Mint. Edited excerpts:

Q. The NEET paper was reportedly leaked from the setter's side, not during distribution. What's your view on the weakest link in preventing leaks?

It's now clear that the question paper was leaked by the people involved in setting the paper. Unlike in 2024, wherein, as reported by the investigating agencies, the leak happened after the paper was printed. Following that, the Dr Radhakrishnan Committee was set up that gave over 250 recommendations. The NTA (National Testing Agency) implemented many of these, which is why, in the 3 May 2026 exam, the leak did not happen after the paper left NTA premises; it was printed, boxed, and delivered to the exam venue without incident.

This time in 2026, the leak happened earlier, by the question paper setters. The foundation of trust for any competitive examination rests on two things: the question paper setters and the invigilators. If either is compromised, the entire system collapses, and one of them is exactly what happened in 2026. The biggest challenge before the task force now is addressing these two points immediately, since the next exam season begins in December 2026.

Q. Has the new task force been given a timeline for its recommendations?

The first meeting will happen in August. We need to suggest short-term, medium-term, and long-term horizons, and it may take about two months to produce an initial framework. I believe that the committee will ultimately agree on a framework applicable not just to one exam, but to the entire examination process—one that ensures exams are conducted in a robust, transparent and risk-proof manner.

Q. What other technology can be used in exams such as these?

A great deal of technology has already gone into examinations. When I took JEE, it was entirely paper and pen. From there, we moved to OMR sheets read by optical scanners, then image scanners, and eventually to computer-based testing. Today, we use multiple layers of technology—biometric verification, encryption, handwriting detectors, and detectors for cell phone activity, among others—to ensure exams are conducted securely. The task force will take stock of all of this and recommend additional measures where necessary, including around the punishment for those found responsible.

Q. When a leak happens, it is usually students who pay the price— cancelled exams, retests, lost time...

It is very important that the people trying to play with the future of children are put behind bars quickly.

Q. How will AI be used to monitor invigilators? Doesn't that raise privacy concerns?

Not necessarily. Think of it like a surveillance camera at a railway station. We need it for security, and we're not capturing anyone's personal information. What we do is set up surveillance cameras in exam rooms feeding into control centres, where every camera is monitored for anything untoward. In 2026, both the May and June exam sessions involved significant effort of this kind. AI can help further—a human watching a screen for two hours might miss a blink-and-you-miss-it moment, but AI can monitor continuously and flag specific incidents, such as two people talking, based on the parameters we define.

Q. Could the new framework include multiple-session testing, like JEE? Is computer-based testing realistic for an exam as large as NEET, with centres also in very small towns?

JEE Mains is conducted for 1.3-1.5 million students, but across multiple sessions. As a country, we can confidently test around 300,000 students at a time—running four to five sessions, twice a weekend, adding up to 10-15 sessions a year. NEET, for various reasons, has had to be conducted as a single-session exam, which is why the NTA hasn't been able to move it to CBT (computer-based test).

Given everything that's happened, one possibility going forward is moving NEET to multiple sessions as well—likely at least eight to ten, given the roughly 2.2 million aspirants involved. That could be a very important shift.

Q. How will you ensure there's no leakage from the setter's side? Since it's a small, select group of people, what can realistically be done?

One of the least talked-about but most important factors here is trust; society itself runs on trust. That's the real challenge. All our encryption and biometric safeguards apply after the question paper is finalized. But if the person who knows the questions chooses to leak them, what can be done? It's a genuinely difficult problem.

In a population of 150 crore, finding 50 good chemistry teachers, 50 good maths teachers, and equally strong biology teachers isn't hard. But we need a far more rigorous selection process for who gets entrusted with this responsibility. That's the specific problem we're focussed on.

Invigilators are a separate, more containable problem, and we do have solutions for that side.