Instead, their questions would be added to a large central repository from which exam papers would later be generated, significantly reducing the number of individuals with knowledge of the final paper.

The move follows the CBI’s investigation into the NEET-UG 2026 leak, which has so far resulted in the arrest of 13 people, including translators and subject experts accused of leaking different sections of the question paper.
As reported by The Indian Express last week, the nature of the arrests points to the leak occurring during the first phase of the confidential operations (CONOPs), which is the highly restricted paper-setting and translation process introduced after the 2024 controversy.

The arrests of translators allegedly linked to the Physics, Chemistry and Biology sections of the exam prompted officials to re-examine the system’s long-standing assumption that trusted experts can safely be given access to complete sections of a paper.

“If the paper is compromised, then it is not a pen-and-paper problem. It is a problem of system design and structure,” a senior government official said. “We are trying to reduce human interface. There has to be a zero-trust architecture.”
Officials said the proposed model would fundamentally change the role of subject experts. Instead of being invited to set questions for a specific examination, experts would contribute questions to a common question bank without being told where those questions would eventually be used.

“The domain expert should be examination-agnostic. He should not know whether the question is being set for NEET, JEE or any other examination. The objective is to ensure that no single individual has visibility of the final paper,” the official said.

Read | NTA to re-conduct NEET UG 2026 on June 21

According to officials, the agency is exploring the creation of a repository containing thousands of questions across subjects. The final paper would then be assembled from this pool using technology-assisted processes designed to maintain subject balance and difficulty levels.

“We can have a bank of 10,000 questions. The final paper can be built from that bank. Technology can help determine difficulty levels and question distribution. The idea is to minimise the number of eyes through which the paper passes,” the official said.

Translation, another area that came under scrutiny following the arrests in the 2026 leak case, may also undergo significant changes. The shift is likely to build on commitments already made by the NTA before the Supreme Court. In an affidavit filed earlier, the agency said it planned to use artificial intelligence for nearly 85 per cent of the translation process, with human experts carrying out the remaining review and validation.

Officials said the agency is now examining how that approach can be integrated into the broader redesign. The objective is not to eliminate human translators but to reduce the amount of time they spend handling sensitive material and limit their visibility of the final paper.

“AI can do the first translation. The human translator’s role should be limited to checking whether the meaning has been correctly conveyed,” the official said. “Even then, we want to ensure that translators are not necessarily aware of the specific examination for which the material is being prepared. They should be validating questions, not handling complete question papers.”

The push for structural reform comes as the NTA prepares for the NEET-UG re-test scheduled for June 21. Officials said some changes have already been introduced, including the use of a fresh set of subject experts. Discussions are also underway on strengthening the transportation and storage of printed papers after they leave the paper-setting stage.

“Some of the people arrested had been associated with the system for years,” the official said. “That is precisely why the system itself has to trust individuals less and processes more. There is an old question, “if the bodyguards themselves become the threat, who protects you?”