How AI cracked a ₹70,000 crore tax evasion trail in Hyderabad’s biryani chains

A sweeping investigation that started with biryani restaurant chains in Hyderabad has uncovered what could be one of the largest tax evasion rackets in India’s food and beverage sector. Using advanced analytics and artificial intelligence tools, the Income Tax Department’s investigation unit in Hyderabad examined nearly 60 terabytes of transaction records from a widely used restaurant billing software operating across the country. The platform, used by over one lakh restaurants and accounting for roughly 10% of the restaurant billing software market, became the focal point of the investigation.

Officials discovered that since the 2019–20 financial year, restaurants using the software allegedly concealed sales worth at least ₹70,000 crore. The final tax liability, including penalties, is still being assessed. The data sweep covered 1.77 lakh restaurant identification numbers nationwide. Investigators identified post-billing deletions amounting to ₹13,317 crore within the broader ₹70,000 crore suppression figure. Andhra Pradesh and Telangana alone accounted for ₹5,141 crore in hidden turnover.

Also Read: THIS US tech CEO missed AI Summit gala hosted by PM Modi after four hours in Delhi traffic. Who is she?

To validate digital findings, authorities carried out detailed inspections, both physical and electronic, at 40 restaurants across Andhra Pradesh and Telangana. These checks revealed approximately ₹400 crore in undisclosed sales. Five states emerged as major hotspots for suspected evasion: Tamil Nadu, Karnataka, Telangana, Maharashtra, and Gujarat. Karnataka recorded the highest volume of deleted transactions at around ₹2,000 crore, followed by Telangana with ₹1,500 crore and Tamil Nadu at ₹1,200 crore. In several cases, officials noted that establishments failed to even erase records, despite under-reporting revenues in tax filings.

Based on sampled data, investigators estimate that nearly 27% of total sales were deliberately left out of official disclosures. The investigation was conducted using data sourced from the software provider’s facility in Ahmedabad. Detailed analysis was carried out at the department’s digital forensic and analytics laboratory in Hyderabad’s Ayakar Bhavan. Authorities uncovered multiple tactics used to suppress revenue. While restaurants typically log all payment modes, cash, card, and UPI, to prevent internal fraud, certain operators selectively removed cash invoices from the system. In other instances, bulk deletions were carried out for specific date ranges, sometimes erasing up to a month’s billing data before filing tax returns that reflected only a fraction of actual earnings.

The dataset reviewed spans six financial years, from 2019–20 to 2025–26, covering billing transactions worth approximately ₹2.43 lakh crore. Investigators deployed high-capacity computing systems and AI tools, including Generative AI, to rapidly map GST numbers to corresponding restaurants using publicly available information and open-source databases.

The inquiry initially focused on Hyderabad, Visakhapatnam, and other towns in Telangana and Andhra Pradesh. After preliminary findings suggested systemic misuse of the billing software, the Central Board of Direct Taxes expanded the operation nationwide. Officials believe the current findings may represent only a fraction of the overall problem, noting that several other billing software platforms operate in the sector and could be subjected to similar scrutiny.