NEET 2026 Cancelled After 140 Questions Leaked — AI and Blockchain Can Fix This

On May 3, 2026, more than 2.27 million students across India sat down to take NEET-UG — the single examination that determines who becomes a doctor in the world's most populous country. Nine days later, the National Testing Agency cancelled the entire exam. Investigators found that up to 140 questions from the actual question paper had been circulating beforehand. Every one of those students — most of whom had spent two to four years and ₹5 to ₹8 lakh preparing — was told to come back and try again on June 21.

This is the second consecutive year NEET has been at the centre of a major paper leak controversy. In 2024, 67 students from a handful of centres scored a perfect 720 out of 720 in an exam that the rest of the country found extremely difficult. The Supreme Court of India intervened. The CBI was called in. The NTA chairman was removed from his post.

And now, in 2026, it happened again.

This Is Not India's Problem Alone

Before discussing solutions, it is important to state clearly: examination fraud is a global crisis, not an Indian one.

In June 2025, question papers from three Cambridge AS and A Level examinations were partially leaked across Pakistan, affecting students whose university applications abroad depended on those results. Cambridge Assessment acknowledged the breach.

In December 2025, two senior officials from South Africa's Department of Basic Education were suspended after mathematics and physics papers for the national matriculation examination reached students at seven schools in Pretoria, in the Tshwane district.

Nigeria's Joint Admissions and Matriculation Board has faced recurring paper leak incidents for over a decade. Kenya's national examination body, KNEC, cancelled multiple papers in 2015 and has faced ongoing challenges since. In 2018, the UK's Edexcel experienced a paper leak affecting GCSE students.

The pattern is global and consistent: a physical exam paper exists somewhere before the exam day, multiple human hands touch it, financial incentives to leak are high, and monitoring systems are not good enough to stop the breach or identify it in time.

The technology to solve this problem exists today and is already being used by leading universities and testing organisations worldwide. The gap is not innovation. The gap is adoption.

The Human and Financial Cost Nobody Calculates

The statistics matter. The individual stories behind them matter more.

A student from a Tier-3 town in India preparing for NEET does not simply study. She reorganises her entire life. She moves to a coaching hub — Kota, Patna, Hyderabad — at 16 or 17 years old. Her parents take loans. They stop spending on anything that is not essential. Her siblings go without things so that the family's best hope can focus on one exam.

When that exam is cancelled nine days after she sat it, the damage is not only emotional, though the emotional damage is severe. The financial cost resets immediately. Another year of coaching fees. Another year of accommodation. Another year of her parents' sacrifice. For families already at the edge of what they can afford, a second year may simply not be possible. Many students in this situation abandon their preparation permanently, not because they lack ability but because their families cannot carry the cost of a third attempt.

Mental health professionals working with competitive exam students across South Asia and sub-Saharan Africa describe a specific psychological state that repeated cancellations produce: a loss of faith not in one's own ability but in the fairness of the system itself. That loss of faith is not easy to restore. It affects motivation, performance, and long-term relationship with education.

The students who suffer most from exam fraud are the honest ones. That is the core injustice this crisis represents.

Why the Same System Keeps Failing Everywhere

Every national examination system built around physical paper has the same structural vulnerability: the paper must travel.

A question paper for a large national examination passes through a chain of people from the moment it is finalised by examiners to the moment a student reads it. That chain includes printing staff, packaging teams, transport personnel, storage facility managers, district officials, and centre coordinators. In a country the size of India or Nigeria, this chain can include 40 to 60 individuals spread across hundreds of kilometres.

Every person in that chain is a potential point of failure. Not because people are inherently dishonest, but because the financial incentive is enormous, the monitoring is inadequate, and the punishment has historically been inconsistent. A leaked paper for a national medical entrance exam can sell for ₹10 to ₹30 lakh on organised black markets. For a low-paid government employee handling examination materials, the temptation is real and the probability of being caught has been low.

This is a system designed before digital technology existed. It is still largely running on trust, tamper-evident seals, and physical logistics that were never designed to be secure against organised crime.

How AI Can Detect Fraud Before and After the Exam

Artificial intelligence cannot make human beings honest. But it can dramatically narrow the window in which dishonesty goes undetected — and it can flag anomalies that no human reviewer would catch at scale.

Pre-exam anomaly detection analyses registration and preparation patterns across examination cohorts to identify statistical outliers before a single paper is distributed. When a particular coaching centre's students show a sudden surge in mock test performance across a narrow window of topics that happen to match the actual paper, an AI system monitoring those patterns can flag this for investigation weeks before exam day.

Post-exam statistical fraud detection is already in use by major international testing organisations including Educational Testing Service in the United States (which runs GRE, TOEFL, and other global examinations) and Cambridge Assessment. These systems run machine learning models across result distributions to identify clusters of suspiciously similar answer patterns, abnormal score jumps at specific centres, and geographic anomalies where one district far outperforms its historical baseline.

If India's NTA had deployed this kind of system in 2024, the 67 perfect scorers from a handful of centres would have triggered an automated flag before results were even declared. The 2026 controversy — where 140 questions matched circulated material — would have been detectable through comparison of circulated paper images against the actual question bank, using image matching and NLP similarity analysis, within hours of the breach rather than days.

Karnataka's Kannada Examinations Authority is currently piloting an AI-based malpractice detection system for competitive examinations, the first Indian state body to formally adopt this approach. It is a model that should scale nationally.

Secure Digital Infrastructure: Remove the Paper, Remove the Risk

The most direct solution to physical paper leaks is to make physical paper unnecessary.

Server-side encrypted examination delivery works as follows: question papers are stored on secure, air-gapped servers and encrypted using AES-256 standards, the same cryptographic standard used by Indian banking infrastructure and US federal government systems. At the moment the examination begins, the paper is decrypted and delivered directly to the candidate's terminal. No complete physical copy of the paper exists anywhere until exam time begins.

This makes it structurally impossible to leak the night before. There is nothing to steal.

This is not theoretical. India's GATE examination — the Graduate Aptitude Test in Engineering, which admits candidates to IITs and NITs — has used computer-based testing with randomised question delivery for years. IBPS, the banking recruitment board, runs large-scale computer-based examinations across India. NTA itself has announced that NEET will move to computer-based format beginning 2027. The technology is proven. The question is the pace of rollout.

For universities abroad currently evaluating examination security: server-side delivery with per-candidate question randomisation from large question banks is the gold standard. No two candidates see questions in the same order. Coordinated leaking becomes essentially pointless because the leaked sequence does not match any individual candidate's actual paper.

Finland's national examination system uses an open-source platform called Abitti, developed through a government-academic partnership, which functions offline and syncs results securely to handle connectivity gaps. This model is directly applicable to regions with unreliable internet — including rural India, rural Nigeria, and rural South Africa — and deserves serious study by examination bodies in these regions.

Blockchain for an Unbreakable Audit Trail

Blockchain technology creates records that cannot be altered after the fact. Applied to examination security, every step in the question paper chain — who set it, who accessed it, when it was transmitted, when it was opened at each centre — is logged in a decentralised ledger that no single administrator can edit or delete.

If a paper is accessed by someone who should not have accessed it, the blockchain record makes this detectable immediately. If a result is altered after declaration, the blockchain record shows the original. This converts what is currently a situation where a breach may take days or weeks to trace — if it is ever traced at all — into one where the chain of evidence is complete and tamper-proof from the moment the paper is created.

Leading universities globally are already implementing blockchain for academic credential verification. The global digital credentials market is on track to reach $1.13 billion by 2026, growing at 21.7% annually. MIT, the University of Melbourne, and the Malta government's entire education credentialling system are among the early adopters.

For examination security specifically, most institutions implementing blockchain use a hybrid architecture: Ethereum (a public blockchain) for transparency and third-party auditability, combined with Hyperledger Fabric (a private, permissioned blockchain) for confidentiality of specific exam content. This allows independent verification that no tampering occurred while protecting the actual question content.

Biometric Verification: Closing the Impersonation Gap

Paper leak is one form of examination fraud. Proxy examination — where a different, more capable person sits the exam in place of the actual candidate — is another, and one that biometric verification addresses directly.

Convolutional neural networks trained on facial recognition can verify candidate identity at examination check-in against Aadhaar-linked biometric records in India, or national ID systems in other countries, with accuracy rates exceeding 99.7% in controlled conditions. Fingerprint or iris verification as a second factor reduces the residual risk of facial recognition edge cases.

For online examinations, AI-powered remote proctoring systems monitor candidates throughout the exam for: multiple faces in the camera frame, eye movement patterns inconsistent with reading on-screen content, audio signatures suggesting external communication, and unusual pauses or navigation patterns. These systems flag suspicious events for human review rather than making automated decisions, which balances security with fairness.

The important safeguard: biometric data must be encrypted at rest and in transit using the same AES-256 standards as the question papers, with strict data retention limits. Examination security must not become a surveillance apparatus that outlasts its purpose.

What Universities Abroad Can Implement Today

For university administrators and examination boards reading this outside India, the practical implementation hierarchy is as follows.

Immediate (under 6 months, software only): Deploy AI anomaly detection on your existing result data. Run similarity analysis on answer patterns from the current sitting against historical performance baselines. Flag any centre or candidate cluster that sits more than three standard deviations outside expected distribution. This requires no new physical infrastructure and can be integrated into existing examination management systems.

Short-term (6 to 18 months): Implement server-side encrypted question delivery for all examinations where computer-based testing is logistically feasible. Transition from sending physical papers to centres to decrypting digital papers at the moment of examination start. Add biometric candidate verification at check-in using camera-based facial recognition cross-referenced against ID records.

Medium-term (18 months to 3 years): Build a blockchain audit trail for all examination data from question setting to result declaration. Adopt a hybrid Ethereum-Hyperledger architecture for the combination of transparency and confidentiality this requires. Integrate with national credential verification frameworks as they develop.

Policy requirement at every stage: Reduce the number of human hands that touch examination materials. Every additional person in the chain is an additional point of vulnerability. Digital delivery does not just protect the paper — it creates a complete, timestamped access log that makes investigation of any breach orders of magnitude faster and more conclusive.

Our Analysis: The 2027 Window Is India's Real Opportunity

The NTA has confirmed that NEET will transition to computer-based examination format beginning 2027. This is the most consequential examination security policy decision India has made in a generation, and it must be executed correctly or it risks reproducing the same structural failures in a digital format.

A computer-based exam that still transmits complete question banks to local centres in advance — rather than using server-side decryption at exam time — does not solve the paper leak problem. It just changes the format of the leak.

The 2027 transition must be designed around three non-negotiable principles: server-side question delivery with per-candidate randomisation, end-to-end encryption of all examination data using banking-grade standards, and blockchain audit trails covering every access event. If those three principles are built into the architecture from the start, the structural vulnerability that has cancelled NEET twice in two years disappears.

The re-exam on June 21, 2026 will be taken on paper, under the same physical security model that has failed twice. The 2.27 million students sitting it deserve better. So does every student who will sit NEET after them.

Key Takeaways

• NEET 2026 cancelled May 12 for 2.27 million students after investigators found 140 questions overlapping with pre-circulated material; re-exam scheduled June 21, 2026
• This is a global problem: Cambridge A Level papers leaked in Pakistan (June 2025); South Africa matric maths and physics papers leaked in Pretoria (December 2025); Nigeria, Kenya, and the UK have all faced similar incidents
• The root cause is structural: physical papers travel through 40 to 60 human hands from creation to exam day — every person is a potential point of failure given financial incentives for leaking that reach ₹30 lakh per paper
• AI anomaly detection can flag suspicious score clusters before results are declared — ETS and Cambridge Assessment already use these systems; Indian state Karnataka's KEA has begun piloting this approach
• Server-side AES-256 encrypted question delivery makes pre-exam leaks structurally impossible — no complete paper exists in physical or transmissible form before exam time begins; GATE already uses this model; NEET transitions to CBT in 2027
• Blockchain audit trail (Ethereum + Hyperledger hybrid) creates a tamper-proof access log from question creation to result declaration; the global digital credentials market reaches $1.13 billion in 2026 at 21.7% CAGR as universities adopt this globally
• Biometric verification using CNNs with AES-256 encrypted biometric data eliminates proxy examination fraud; accuracy exceeds 99.7% in controlled testing environments
• Universities abroad can deploy AI anomaly detection within 6 months using existing result data, with no new physical infrastructure required; server-side digital delivery and blockchain audit trails follow on an 18-month horizon

Sources

• 2026 NEET Controversy — Wikipedia
• NTA Dismisses Re-NEET Paper Leak Rumours Ahead of June 21 Exam — Outlook India
• NEET UG 2026 Re-Exam: NTA Launches Portal to Report Fake Websites — India TV News
• Cambridge Says Aware of Reports About Question Paper Leak — Dawn Pakistan
• Two South Africa Officials Suspended Over Matric Exam Paper Leaks — AllAfrica
• Secure Online Examination with Biometric Authentication and Blockchain Framework — Hindawi
• Blockchain Credentials Transform Higher Education — EveryCred
• How to Prevent Question Paper Leaks — Eklavvya
• KEA to Utilise AI to Prevent Malpractice in Competitive Exams — Deccan Herald
• Our coverage: Iran Peace Deal Unblocks Gulf AI Buildout