Initially the authorities planned to crackdown only on three reputable biryani chains in Hyderabad. However, it eventually led to one of the biggest tax evasion probes in the Indian food and beverage industry, resulting in the unearthing of a massive ₹70,000 crore suppressed turnover scam involving more than 1.7 lakh restaurants nationwide and going back to the 2019-20 fiscal year.
This matter has been a turning point for AI-powered tax enforcement. It showcases how generative artificial intelligence and big data analytics have the potential to detect deeply hidden and systemic fraud under layers of routine restaurant billing.
The IT Raids on Biryani Chains
On November 18 and 19, 2025, Income Tax officials conducted raids in Hyderabad targeting the well-established food chains named Pista House, Shah Ghouse Cafe and Mehfil.
The Income Tax Investigation Wing conducted simultaneous raids at various locations of these food chains and, in addition to locating company offices, major outlets, and staff quarters, they managed to reach the residences of owners, directors, and senior employees as well. They also scrutinized sales ledgers, branch, wise accounts, digital records, and financial documents to confirm any suspected discrepancies in reported income. These suspicions had been aroused by indications such as the existence of suppressed turnover, unaccounted cash flow, and mismatches between declared and actual earnings.
During the raid, about ₹6 crore in unaccounted cash was allegedly recovered. The department has also looked into the allegations that the cash earnings were diverted through UPI accounts of employees and that there were discrepancies between the reported and actual earnings, especially during the busy business seasons like Ramzan.
Who Are These Biryani-Chains?
All the three brands are not only familiar faces in Hyderabad but are also known outside the city.
Pista House is probably the most well, known haleem and biryani brand in Hyderabad. Its 44-outlet chain and high season sales of haleem have attracted the attention of the IT department, which has sent four teams to the residence of the owner Mohammad Abdul Mazeed alone. The chain has operations in India, the Middle East, and the US.
Shah Ghouse, a name synonymous with Hyderabad biryani was started in 1973 by Mohammed Ghouse Pasha with the help of Mohammed Rabbani and Mohammed Irfan. The chain now has several food outlets all over Hyderabad and dominates the delivery market.
Mehfil Group, which is a 15-outlet brand spread across Hyderabad, was set up by Jaleel Rooz and Mubeen Pasha in 2006. Besides a branch in the UAE, Mehfil’s quick growth and overseas expansion are said to be the reasons for the financial investigations.
The Billing Software Trail: 60 Terabytes of Data
The Hyderabad raids were merely the starting point. The results the police uncovered by tracing the digital footprints were even more terrifying.
Officials came across concrete evidence of a large-scale sales suppression scam in the restaurant sector throughout India. They went through approximately 60 terabytes of billing data extracted from a leading restaurant billing software, which is used by over one lakh restaurants across the country, a platform that is linked to around 1.77 lakh restaurant IDs.
That software was the tool that facilitated the fraud. Once customers had paid, bills were either deleted or modified in such a way that no tax inspector would ever find any clues, or at least that’s what the restaurateurs thought.
Officials with the help of cutting-edge data analytics and AI tools, including Generative AI, reviewed transactions from 2019-20 to 2025-26. The total amount of billing entered in the system during this time was approximately ₹2.43 lakh crore. After comprehensive analysis and testing, the officials believe almost 27% of the sales were hidden.
One of the main investigative methods used was the “biryani method“: since the price of biryani is fixed and the amount of ingredients for one serving is known, officials could calculate the approximate sales of a restaurant just by looking at their raw material purchases, then compare that with declared turnover.
The Scale of Evasion: State by State
The Income Tax Department’s Hyderabad investigation unit has found that since the 2019-20 financial year, restaurants have hidden sales to the tune of at least ₹70,000 crore. Some restaurants that used billing software were found to have deleted bills to the extent of more than ₹13,317 crore. In just two states, Andhra Pradesh and Telangana, the hidden sales were worth ₹5,141 crore.
The top five states where tax evasion was discovered were Tamil Nadu, Karnataka, Telangana, Maharashtra, and Gujarat. Karnataka had the greatest number of deleted transactions, which were worth roughly ₹2,000 crore. Next was Telangana with ₹1,500 crore and Tamil Nadu with ₹1,200 crore.
Out of the total 3,734 PANs in Telangana and Andhra Pradesh, sales suppression was found in 2,650 PANs. Out of these, 684 cases showed sales suppression of over ₹1 crore, and 416 of these were from Hyderabad only. In 231 instances, zero or no GST turnover was reported, whereas in 155 cases the declared turnover was less than 1/12th of the suppressed turnover.
The physical checks corroborated the digital evidence. The officers carried out both physical in-depth and digital verification at 40 restaurants in Andhra Pradesh and Telangana and identified a sales suppression of ₹400 crore.
How AI Cracked the Code
The breakthrough in the investigation represents a major turning point in the way tax authorities in India enforce laws. Authorities were able to rapidly identify discrepancies by first mapping GST numbers to restaurants through the use of open-source and publicly available online information. Data from delivery apps, GST filings, and internal software logs were cross, checked to spot inconsistencies that no manual audit would have been able to uncover at this scale.
The software is used by about one out of ten restaurant billing software accounts in India, so the total problem could be many times larger if all platforms are considered. The officials think that this is just a start and that there are many other billing platforms scattered across the country.
Tax officers said that the use of generative AI to carefully examine 60 terabytes of transactional data and to detect patterns such as post, payment bill deletion, time, stamp manipulation, and ghost entries is a game-changer in financial forensics.
Industry Practices That Enabled the Fraud
Investigators found that a number of standard industry practices actually made it even easier to hide things. A rapid table turnover, big takeaway orders, and cloud kitchens that operate under several different brand names from one single place are all typical restaurant industry practices. All of these make the process very difficult for the revenue authorities as they make the verification of revenue a historical habit that was not easy to carry out.
Plus with the Cloud Kitchen concept, the management was able to operate multiple virtual brands at one authorized location. Thus, localizing the revenue was made very complicated through multiple GST registrations.
What Comes Next
The inquiry has now gone far beyond Hyderabad. The findings of the IT department have led to the expansion of the investigation on a national level, AI-based audits are said to be transforming the nature of tax enforcement, and this may be just the tip of a much larger problem.
It is anticipated that tax authorities will send out notices to thousands of restaurant operators in the entire country, and reports suggest that the Central Board of Direct Taxes (CBDT) is mulling over a comprehensive sector-specific audit framework for the food and beverage businesses that are using digital billing platforms.
For India’s restaurant industry which is worth hundreds of thousands of crores the message is crystal clear: the days when revenue was concealed in deleted bills are now gone. The algorithms are present, and they don’t miss a single plate of biryani.
