IRCTC flags 9 crore user accounts, expands AI monitoring across railway kitchens

NEW DELHI: Stepping up efforts to address the problem of passengers struggling to book confirmed tickets, IRCTC has deactivated more than three crore suspicious user IDs and also placed another six crore under verification of authenticity to prevent misuse.

The railways’ catering and ticketing arm has also expanded its AI-based kitchen monitoring system across its network to deal with complaints about food hygiene and quality. Officials said more than 800 kitchens are now being monitored through 2,394 AI-enabled cameras that detect nine types of issues that are primarily responsible for unhygienic food in trains.

These include head cap compliance, transparent gloves detection, mopping, wiping, rodents, flies and cockroaches.

On the crackdown on suspicious users booking tickets, officials said 501 complaints have been lodged on the National Cyber Crime Portal pertaining to 4.18 lakh suspicious PNRs. It has also taken action against fraudulent digital identities by blocking 13,343 suspicious email domains during the year.

On monitoring kitchens, a senior IRCTC official said the AI-enabled cameras can detect anything that’s as small as the size of an ant (7-8 mm). IRCTC served about 60 crore meals in 2025-26.

The official added that they are expanding this surveillance as more kitchens are added to its network. The cameras supported by AI tools quickly identify unhygienic practices and send alerts.

“Any issue flagged is sent to the kitchen manager concerned and it escalates, if the issue is not resolved. Action is taken against the person responsible within two hours, and the error is rectified,” the official supervising the IRCTC war room said.

Failure to wear head caps is the most common hygiene-related violation detected by the AI system in kitchens. On average, the system generates around 350 alerts or error tickets daily.

Officials said the most difficult period is during seasonal weather changes when the quality of food can get affected. Maximum number of complaints are received during this period. They added that while the AI system generates around 350 tickets a day, the effective error rate is about 10%.